Association of obstructive sleep apnea with arterial stiffness and nondipping blood pressure in patients with hypertension

Abstract

Whether sex influences the association of obstructive sleep apnea (OSA) with markers of cardiovascular risk in patients with hypertension is unknown. In this study, 95 hypertensive participants underwent carotid‐femoral pulse wave velocity, 24‐hour ambulatory blood pressure monitoring, echocardiogram, and polysomnography after a 30‐day standardized treatment with hydrochlorothiazide plus enalapril or losartan. OSA was present in 52 patients. Compared with non‐OSA patients, pulse wave velocity values were higher in the OSA group (men: 11.1±2.2 vs 12.7±2.4 m/s, P=.04; women: 11.8±2.4 vs 13.2±2.2 m/s, P=.03). The proportion of diastolic dysfunction was significant in men and women with OSA. Compared with non‐OSA patients, nondipping systolic blood pressure in OSA was higher in men (14.3% vs 46.4%) and in women (41.4% vs 65.2%). OSA was independently associated with pulse wave velocity (β=1.050; P=.025) and nondipping systolic blood pressure (odds ratio, 3.03; 95% confidence interval, 1.08–8.55; P=.035) in the regression analysis. In conclusion, OSA is independently associated with arterial stiffness and nondipping blood pressure in patients with hypertension regardless of sex.

1. Introduction

Obstructive sleep apnea (OSA) is a common condition characterized by recurrent episodes of upper airway obstruction during sleep, resulting in reductions of intrathoracic pressure, intermittent hypoxia, and sleep fragmentation.1 Traditional risk factors for OSA include male sex, age, and obesity.2, 3 Despite the higher predominance of OSA in men (2:1 in population‐based studies and up to 8:1 in sleep laboratory–based populations),3, 4, 5 OSA is more commonly underdiagnosed in women.6 Several atypical symptoms, including fatigue, lack of energy, and tiredness, have been described and are more common in women with OSA,7 and they may partially explain the underdiagnosis of OSA in this population.

Growing evidence suggests that OSA is associated with increased surrogate markers of cardiovascular risk8, 9 and cardiovascular events, such as myocardial infarction10, 11 and stroke.12, 13 However, most of the evidence is based on men. For example, we previously described that OSA in men was associated with increased arterial stiffness and heart remodeling, especially in patients with hypertension.8 Moreover, one investigation (74% men) showed that OSA is frequently associated with nondipping status in ambulatory blood pressure (BP) monitoring.14 All of this evidence supports the notion that OSA is an emerging cardiovascular risk factor, but it is important to stress that the observed higher risk associated with OSA may not necessarily be true for women.

Because the cardiovascular impact of OSA in women is poorly understood, we evaluated whether sex potentially influences surrogate markers of cardiovascular risk in OSA patients with hypertension. Based on the measurements of arterial stiffness, diastolic dysfunction, and BP, we hypothesized that women with OSA have similar impairment to men with an established diagnosis of hypertension.

2. Methods

We recruited consecutive patients from the outpatient hypertension clinic at the Heart Institute, University of São Paulo Medical School, Brazil, between January 2010 and December 2012. All patients had grade 2 or 3 hypertension (office BP >160×100 mm Hg)15 at admission. We excluded patients with secondary forms of hypertension (except OSA), creatinine levels >2.0 mg/dL, and a low life expectancy, as well as patients who had a stroke, acute coronary syndrome, cardiac failure, or peripheral arterial disease.

After recruitment, all patients were given standardized antihypertensive treatment with a thiazide diuretic (hydrochlorothiazide 25 mg) plus enalapril (20 mg BID) or losartan (50 mg BID) in case of enalapril intolerance (such as persistent cough). This procedure is routinely performed and based on current guidelines.15 During this period, all patients were followed by a pharmacist who monitored treatment adherence (>70%) through pill counting. After 30 days, all volunteers underwent the following procedures, which were performed by examiners blinded to OSA status.

2.1. Clinical evaluation

We collected medical history and anthropometric measurements, including neck and waist circumferences and body mass index. Office BP was determined after the patient rested in a sitting position for three or more readings (the first reading was excluded) obtained at 5‐minute intervals with an automatic and validated digital machine (Omron 742; Kyoto, Japan). The use of cigarettes and alcohol was not allowed on this clinical evaluation day. In addition, we administered sleep questionnaires to evaluate the risk of OSA (Berlin Questionnaire)16 and excessive daytime sleepiness via the Epworth Sleepiness Scale.17

2.2. Arterial stiffness

For the measurement of arterial stiffness, a single researcher measured the carotid‐femoral pulse wave velocity (PWV) using Complior (Colson, Garges les Genosse, France) as previously described.18 Briefly, all examinations were performed in the morning after the patients had been lying in a supine position for at least 10 minutes in a quiet environment with a controlled and stable temperature. The common carotid artery and femoral artery pressure waveforms were noninvasively recorded using a TY‐306 Fukuda pressure‐sensitive transducer (Fukuda, Tokyo, Japan). The pressure waveforms were digitized at a sample acquisition frequency of 500 Hz. The two pressure waveforms were then stored in a memory buffer. A preprocessing system automatically analyzed the gain in each waveform and adjusted it to equalize the two signals. When the operator observed a pulse waveform of sufficient quality on the computer screen, digitization was suspended and calculation of the time delay between the two pressure upstrokes was initiated. Measurements were repeated over 10 different cardiac cycles, and the mean was used for the final analysis. The distance traveled by the pulse wave was measured over the body surface as the distance between the two recording sites (D), whereas pulse transit time (t) was automatically determined by the Complior device. PWV was automatically calculated as PWV=D/t.

2.3. Ambulatory BP monitoring

We performed 24‐hour ambulatory BP monitoring (Spacelabs model 90207, Spacelabs Healthcare, Snoqualmie, WA, USA). Appropriate cuff sizes were used for all patients. We considered a minimum of 14 successful measurements during the day and 7 at night.19 BP was measured every 10 minutes during the day and every 20 minutes during the night with an appropriate cuff placed on a nondominant arm. Activity, bedtime, and time on awakening from sleep were recorded by participants in diaries. Participants were instructed to perform their ordinary daily activities and not to move their arm during the ongoing measurement. Values were considered normal when the systolic and diastolic BPs were <135 mm Hg and 85 mm Hg during the daytime and <120 mm Hg and 70 mm Hg during the nighttime, respectively.20

2.4. Transthoracic echocardiogram

All examinations were performed using a Sequoia 512 echocardiograph (Acuson, Mountain View, CA, USA) and a 3.6‐1.8–MHz transducer (3V2c) and were copied to a storage device (magneto‐optical disk drive) for analysis. Noninvasive arterial pressure was acquired during the examination. Echocardiographic images were obtained in the parasternal long‐ and short‐axis; apical two‐, three‐, four‐, and five‐chamber; and subcostal views. The American Society of Echocardiography (ASE) guidelines were used during two‐dimensional, M‐mode, Doppler echocardiography and the acquisition and calculation of various parameters.21 These parameters were measured during at least three cardiac cycles. All echocardiogram results were analyzed by the same experienced echocardiographer. A detailed description of the echocardiograph technique can be found in the Appendix S1.

2.5. Blood samples

Fasting blood samples were obtained to measure glucose, lipid profile, urea, and creatinine using standard techniques. Estimated glomerular filtration rates were calculated using the equations from the Chronic Kidney Disease Epidemiology consortium22 without correction for race (has not added useful information for Brazilians).23

2.6. Polysomnography

Beyond the aforementioned evaluation, all patients were submitted to a full overnight polysomnography study (Embla; Flaga hf. Medical Devices, Reykjavik, Iceland). Apnea was defined by a reduction of airway flow of ≤90% for at least 10 seconds. Hypopnea was defined by a reduction of airway flow of at least 50% for 10 seconds associated with arousals or ≥3% oxyhemoglobin desaturation.24 The apnea‐hypopnea index (AHI) was calculated as the total number of respiratory events (apneas plus hypopneas) per hour of sleep. OSA was defined by an AHI ≥15. This cutoff, used by others,25 relies on the fact that recent evidence suggests that mild OSA may not have significant cardiovascular consequences.26 A single experienced observer performed all sleep analyses with no access to the cardiovascular evaluation.

2.7. Statistical analysis

Data were analyzed with SPSS statistical software version 20.0 (IBM Corp, Armonk, NY, USA). Quantitative variables are expressed as means±standard deviations. We performed two‐way analysis of variance and generalized linear model to compare the four groups (men and women with and without OSA). After checking normality with the Kolmogorov‐Smirnov test, linear regression models were created to evaluate the influence of sex and OSA on PWV. In addition, logistic regression analyses were performed to evaluate the independent predictors of the presence of diastolic dysfunction and nondipping status. Potential confounding factors, such as age, body mass index, office mean BP, diabetes and smoking (past/current), and estimated glomerular filtration rate (for PWV analysis) were included in these models. We also performed additional analysis evaluating the impact of using mean BP derived from ambulatory BP monitoring instead of office mean BP on PWV. All tests were performed at a 5% significance level.

3. Results

We initially selected 125 patients. Thirty of these patients were excluded (the reasons for excluding patients are presented in Figure 1). Thus, 95 patients were included in the final analysis (53 women [56%]). OSA was present in 52 patients (54.7% [men: 66.7%; women: 45.3%] (P=.041). Table 1 shows the main clinical data. Compared with the patients without OSA, the OSA group was older and had a higher frequency of men. Moreover, body mass index, neck and waist circumferences, and proportion of systolic BP dipping were higher in the OSA group. The estimated glomerular filtration rate was lower in patients with OSA. As expected for the study design, patients with OSA presented higher AHI. Higher percentage of stage 1 nonrapid eye movement sleep stage was also observed as a consequence of sleep fragmentation induced by obstructive events in both sexes. Table S2 shows the reported use of antihypertensive medications stratified by OSA before starting the current protocol under standardized medications. There were no differences in medications according to the presence of OSA.

Figure 1.

Flowchart detailing patient recruitment. *All measurements were performed within a 30‐day period. Clinical evaluation, office blood pressure (BP), pulse wave velocity (PWV), and echocardiogram were performed on the same day. Polysomnography and ambulatory blood pressure monitoring (ABPM) were performed on distinct days. OSA indicates obstructive sleep apnea

Table 1.

Clinical Data According to the Presence of OSA After Standard Antihypertensive Treatment

Variable	All Patients (N=95)	No OSA (n=43)	OSA (n=52)	P Value
Age, y	56±10	53±9	59±9	.005
Men, %	44	32.5	53.8	.03
BMI, kg/m2	30.4±5	29±4.7	31.6±4.8	.01
Neck circumference, cm	39.1±4.0	37.7±4.0	40.3±3.6	.002
Waist circumference, cm	101.8±11.2	97.5±10.5	105.3±10.5	<.001
Diabetes, %	24.7	26.1	23.5	.47
Dyslipidemia, %	51.6	52.3	51.0	.52
Current smoking, %	16.1	16.6	15.6	1.0
Hypothyroidism, %	3.2	2.3	4	1.0
Hypertension diagnosis, y	13.4±10	14±11	12.8±9.5	.59
eGFR, mL/min/1.73 m2	85±20	91±19	81±21	.03
Office SBP, mm Hga	153±23	150±22	155±23	.32
Office DBP, mm Hga	96±14	97±13	95±14	.6
Ambulatory BP monitoring
SBP, 24‐h, mm Hga	131±17	129±18	133±16	.24
SBP, wake, mm Hga	135±17	133±17	137±16	.23
SBP, sleep, mm Hga	121±18	117±18	124±17	.1
SBP, nondipping, %	44.7	32.6	55	.02
DBP, 24‐h, mm Hg	81±12	80±12	81±12	.69
DBP, wake, mm Hg	84±12	84±12	84±12	.78
DBP, sleep, mm Hg	71±13	70±13	72±13	.45
DBP, nondipping, %	27.7	30	25,5	.44
Polysomnography data
Total sleep time, min	333.5±77.5	336±74	331.5±81	.78
N1, %	14±8	11±5	16±9	.001
N2, %	52±8	54±8	51±8	.07
N3, %	19±8	20±7	19±8	.42
REM, %	15±6	15±6	15±6	.79
AHI, events per h	22±19	7±5	34±17	<.001
Baseline SpO2, %	97±2	98±2	97±2	.07
Lowest SpO2, %	82±2	87±6	79±8	<.001
Total sleep time SpO2 <90%, %	0.4(0.03–2.7)	0.02(0–0.3)	1.4(0.3–5.5)	<.001

Abbreviations: AHI, apnea‐hypopnea index; BMI, body mass index; BP, blood pressure; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; N1, stage 1 of sleep; N2, stage 2 of sleep; N3, stage 3 of sleep; OSA, obstructive sleep apnea; REM, rapid eye movement; SBP, systolic blood pressure; SpO₂, oxyhemoglobin saturation. P values in bold indicate statistical significance. Sleep stages are reported as a percentage of total sleep time.

^aObtained after 1 month of standardized antihypertensive treatment as described in the methods section.

Table 2 shows the main clinical data stratified by sex. Interestingly, we found a greater prevalence of nondipping pattern in women than in men, despite the lower OSA frequency. Obesity parameters, metabolic syndrome diagnosis, nondipping systolic BP, and percentage of stage 1 nonrapid eye movement sleep stage were significantly different between the OSA and non‐OSA groups regardless of sex. In women, OSA patients were older and had higher body mass index. The estimated glomerular filtration rates were not different between the OSA and non‐OSA groups in both sexes (Table 2).

Table 2.

Clinical Data According to Sex and the Presence of OSA After Standard Antihypertensive Treatment

Characteristics	Men, No OSA (n=14)	Men, OSA (n=28)	Women, No OSA (n=29)	Women, OSA (n=24)	P Sex	P OSA	P Interaction
Age, y	56±8	58±10	52±10	59±9	.623	.016	.212
BMI, kg/m2	29.5±2.5	30.5±4.5	28.7±5.5	32.8±5.0	.489	.016	.148
Neck circumference, cm	42±2	43±2	35.8±3.1	37.5±2.7	<.001	.012	.697
Waist circumference, cm	101±6	107±11	95.5±12	103.2±9.4	.030	.004	.668
Diabetes, %a	21.4	18.5	28.6	29.2	.334	.458	.362
Dyslipidemia, %a	57.1	44.4	50	58.3	.542	.497	.497
Metabolic syndrome, %a	50	70.4	68	91.7	.035	.020	.468
Current smoking, %a	14.3	18.5	17.8	13	.869	.927	.545
Menopause, %b	–	–	60.7	86.5	–	.110	–
eGFR, mL/min/1.73 m2	79±19.2	69.7±17	95.6±17.7	93±18	<.001	.164	.446
Office SBP, mm Hgc	155±26	153±24	148±19	157±23	.859	.447	.249
Office DBP, mm Hgc	99±13	94±13	96±14	97±15	.969	.532	.359
Ambulatory BP monitoring
SBP, 24‐h, mm Hgc	130±17	135±17	128±19	130±15	.305	.344	.672
SBP, awake, mm Hgc	136±17	139±17	131±18	134±15	.126	.412	.972
SBP sleep, mm Hgc	117±16	125±18	118±20	121±16	.708	.118	.484
SBP, nondipping, %a	14.3	46.4	41.4	65.2	.031	.011	.514
DBP, 24‐h, mm Hgc	82±10	84±12	79±13	78±12	.115	.904	.483
DBP, awake, mm Hgc	86±10	87±13	83±12	81±11	.053	.916	.591
DBP, sleep, mm Hgc	71±12	74±12	70±13	70±13	.360	.544	.465
DBP, nondipping, %a	21.4	21.4	34.5	34.8	.180	.989	.989
Polysomnography data
Total sleep time, min	318±83	340±86	344±70	321±75	.863	.986	.189
N1, %	12±4	18±11	10±5.5	13.5±6	.049	.009	.562
N2, %	52±10	50±8	55±7	52.5±7	.121	.167	.987
N3, %	18±7	17±9	21±7.5	21±6.5	.038	.708	.576
REM, %	17±6	16±7	14±6	14±6	.098	.494	.687
AHI, events per h	8.6±4	36±18	6±4.7	31.6±16	.233	<.001	.815
Baseline SpO2, %	97±2	97±2	98±1	97±2	.209	.222	.067
Lowest SpO2, %	86±4	80±8	88±6	78±7	.947	<.001	.172
Total sleep time SpO2 <90%, %	0.03(0–0.43)	1.6(0.3–4.9)	0(0–0.21)	1.1(0.52–6.5)	.096	.163	.474

Abbreviations: AHI, apnea‐hypopnea index; BMI, body mass index; BP, blood pressure; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; N1, stage 1 of sleep; N2, stage 2 of sleep; N3, stage 3 of sleep; OSA, obstructive sleep apnea; REM, rapid eye movement; SBP, systolic blood pressure; SpO₂, oxyhemoglobin saturation. P values in bold indicate statistical significance. Sleep stages are reported as a percentage of total sleep time.

^aGeneral linear model with binomial distribution and logit link function. ^bChi‐square test. ^cObtained after 1 month of standardized antihypertensive treatment as described in the methods section.

Patients with OSA presented higher PWV values than those in the non‐OSA group (12.9±2.3 vs 11.5±2.3 m/s, P=.007). When stratified by sex, both men and women with OSA presented higher PWVs than their non‐OSA counterparts (Figure 2).

Figure 2.

Unadjusted pulse wave velocity measurements according to the presence of obstructive sleep apnea (OSA) and sex

Table 3 shows the echocardiogram results stratified by sex. We noted that OSA was associated with greater signs of atrial and ventricular remodeling and a higher proportion of diastolic dysfunction in both sexes.

Table 3.

Echocardiographic Data According to Sex and the Presence of OSA

Echocardiogram	Men, No OSA (n=14)	Men, OSA (n=28)	Women, No OSA (n=29)	Women, OSA (n=24)	P Sex	P OSA	P Interaction
LA diameter, mm	38.1±3.5	40.5±3.4	35.4 ±3.8	37.3 ± 2.6	<.001	.006	.744
LA volume, mL	34.2±5.1	34.4±7.6	25.1±7.7	29.1±6.6	<.001	.184	.831
LA volume/BSA	17.7±2	17.5±3.5	14.8±4.6	16.4±3.4	.001	.427	.834
LV mass, g	178±28.8	188±52.8	132±28.2	145±30.8	<.001	.263	.346
LV mass/BSA	93±16	95.5±25.4	77.4±15.7	81.7±17.2	.024	.577	.420
LV remodeling, %a	38.4	66.6	33.3	50	.323	.045	.612
E/A	0.9±0.31	0.77±0.25	0.89±0.27	0.71±0.18	.555	.007	.665
DCT	231±36.2	251±58	215±30.5	239±65.8	.232	.057	.848
IVRT	110±17.7	114±15.1	101±15	112±14.8	.121	.043	.361
Diastolic dysfunction, %a	54	79	46.1	81.8	.890	.005	.637

Abbreviations: A, peak late diastolic velocity; BSA, body surface area; DCT, deceleration time; E, peak early diastolic velocity; IVRT, isovolumic relaxation time; LA, left atrium; LV, left ventricle; OSA, obstructive sleep apnea. P values in bold indicate statistical significance.

^aGeneralized linear model with binomial distribution and logit link function.

We performed a linear regression analysis of PWV (Table 4). In an unadjusted model, we found variables associated with PWV, namely, age, mean BP parameters, current smoking, and OSA (P=.006). In the adjusted model (considering all variables presented in Table 4), OSA was independently associated with PWV when we used office mean BP and daytime mean BP. A strong trend was observed for both 24‐hour mean BP (P=.05) and nighttime mean BP (P=.077). The interactions of OSA and sex on PWV were not significant (Table 4).

Table 4.

Linear Regression With PWV as the Dependent Variable

Characteristics	Unadjusted		Adjusted Model 1a		Adjusted Model 2a		Adjusted Model 3a		Adjusted Model 4a
	Coefficient β (SE)	P Value	Coefficient β (SE)	P Value	Coefficient β (SE)	P Value	Coefficient β (SE)	P Value	Coefficient β (SE)	P Value
Age, y	0.122 (0.024)	<.001	0.121 (0.025)	<.001	0.136 (0.022)	<.001	0.138 (0.022)	<.001	0.127 (0.022)	<.001
BMI, kg/m2	–0.025 (0.059)	.668	–0.110 (0.054)	.040	–0.074 (0.046)	.103	–0.077 (0.045)	.091	–0.074 (0.047)	.113
Office mean BP, mm Hg	0.009 (0.017)	.581	0.019 (0.014)	.178	–	–	–	–	–	–
24‐h mean BP, mm Hg	0.064 (0.018)	<.001	–	–	0.076 (0.015)	<.001	–	–	–	–
Daytime mean BP, mm Hg	0.061 (0.019)	.001	–	–	–	–	0.075 (0.015)	<.001	–	–
Nighttime mean BP, mm Hg	0.066 (0.017)	<.001	–	–	–	–	–	–	0.066 (0.015)	<.001
Diabetes	0.028 (0.587)	.962	0.172 (0.515)	.738	0.102 (0.443)	.818	0.108 (0.442)	.807	0.118 (0.454)	.796
Former smoking	0.115 (0.741)	.876	0.611 (0.666)	.359	0.294 (0.565)	.603	0.338 (0.562)	.548	0.202 (0.582)	.728
Current smoking	–1.279 (0.574)	.026	–1.462 (0.498)	.003	–1.135 (0.422)	.007	–1.149 (0.420)	.006	–1.114 (0.432)	.010
OSA (AHI ≥15)	1.392 (0.503)	.006	1.050 (0.468)	.025	0.799 (0.407)	.050	0.859 (0.405)	.034	0.741 (0.419)	.077
Female sex	0.230 (0.521)	.659	0.316 (0.503)	.529	0.696 (0.444)	.117	0.794 (0.444)	.073	0.450 (0.454)	.321
eGFR	–0.025 (0.013)	.051	–0.004 (0.013)	.788	0.011 (0.012)	.382	0.009 (0.012)	.474	0.012 (0.013)	.327

Abbreviations: AHI, apnea‐hypopnea index; BMI, body mass index; eGFR, estimated glomerular filtration rate; OSA, obstructive sleep apnea; PWV, pulse wave velocity; SE, standard error. P values in bold indicate statistical significance.

^aAll variables were included simultaneously in adjusted models with the following differences: adjusted model 1: using office mean blood pressure (BP); adjusted model 2: using 24‐hour mean BP; adjusted model 3: using daytime mean BP; and adjusted model 4: using nighttime mean BP.

We also performed a logistic regression analysis for the presence of diastolic dysfunction and nondipping status. In an unadjusted model, age, PWV, and the presence of OSA were associated with the presence of diastolic dysfunction. In the adjusted model, only age was independently associated with diastolic dysfunction. The independent association between OSA and diastolic dysfunction was statistically borderline (Table S3). Regarding ambulatory BP monitoring, the adjusted model showed that age, OSA, and women were independently associated with nondipping systolic BP. Similarly, the interactions of OSA and sex on nondipping systolic BP were not significant (Table 5).

Table 5.

Logistic Regression for Independent Predictors of the Presence of Nondipping Systolic Blood Pressure in Hypertensive Patients

Characteristics	Unadjusted		Adjusted
	OR (95% CI)	P Value	OR (95% CI)	P Value
Age, y	1.06 (1.01–1.11)	.026	1.06 (1.01–1.12)	.031
BMI, kg/m2	1.02 (0.94–1.11)	.650	0.99 (0.89–1.09)	.804
Diabetes	1.49 (0.6–3.75)	.392	1.69 (0.58–4.95)	.337
Former smoking	0.86 (0.33–2.2)	.746	0.87 (0.3–2.54)	.797
Current smoking	2.1 (0.64–6.85)	.218	3.58 (0.91–14.08)	.067
OSA (AHI ≥15)	2.52 (1.08–5.85)	.032	3.03 (1.08–8.55)	.035
Female sex	1.95 (0.84–4.46)	.118	3.21 (1.18–8.77)	.023

Abbreviations: AHI, apnea‐hypopnea index; BMI, body mass index. Interaction between obstructive sleep apnea (OSA) and sex in the adjusted model: odds ratio (OR), 0.58 (95% confidence interval [CI], 0.07–5.03) (P=.621). P values in bold indicate statistical significance.

4. Discussion

This study was designed to evaluate whether sex may influence the association of OSA with arterial stiffness, diastolic dysfunction, and BP in patients with hypertension. The following new findings were observed: (1) the increased arterial stiffness seen in men with OSA and hypertension is also seen in women; and (2) OSA is independently associated with increased arterial stiffness and nondipping systolic BP, regardless of sex. Together, our results suggest that women with hypertension are not spared from the cardiovascular consequences of OSA.

It is well established that OSA is more common among men than women in the general population, and we found consistent results in patients with hypertension. The perception that OSA is much more common in men may partially explain the relatively little attention paid to investigate the consequences of OSA among women, including the cardiovascular risk. The available literature provides conflicting results regarding the influence of sex on mortality and cardiovascular events.27 The Sleep and Heart Health Study showed that severe OSA was an independent predictor of incident coronary heart disease, but only in 40‐ to 70‐year‐old men but not in women.28 In the same study, severe OSA was a predictor of heart failure in men but not in women. In contrast, a prospective cohort in Spain that followed 1116 women for up to 72 months found that severe OSA was associated with cardiovascular death.29

To date, limited evidence has supported sex differences in surrogate markers of cardiovascular risk in patients with OSA. Faulx and colleagues30 previously reported that endothelial dysfunction evaluated by brachial artery ultrasonography was only impaired in women with moderate to severe OSA. Consistently, another study in a middle‐aged population revealed that OSA was associated with impaired vascular function (evaluated by digital peripheral arterial tonometry) only in women, independent of menopausal status.31 More recently, data from the Atherosclerosis Risk in Communities‐Sleep Heart Health Study cohort showed that OSA was independently associated with higher levels of high‐sensitivity troponin among women but not men. During the mean follow‐up period of 15 years, OSA was associated with incident heart failure or death only in women.32 In contrast to these previous investigations, our study focused on exploring the potential impact of OSA and sex in consecutive patients with hypertension. An important argument for exploring comorbidities in hypertension relies on the fact that high BP is one of the leading risk factors for global burden of disease.33 By studying a validated marker of arterial stiffness that predicts cardiovascular events,34 we found that the arterial stiffness increase observed in OSA patients with hypertension is not sex‐specific. Nocturnal and nondipping BP are also predictors of cardiovascular events, including in patients with hypertension.35, 36 We found a greater frequency of nondipping pattern in women than in men, despite a lower OSA frequency. Although we do not have definitive explanations for this interesting finding, it is possible that other types of sleep disorders, such as poor sleep quality and insomnia, might be more prevalent in women.37 A recent report found that nondipping was a marker for poor prognosis in patients with OSA38 but a lack of a control group (no OSA patients) limited a definitive conclusion on whether nondipping status has more impact in patients with than without OSA. Our finding that the presence of OSA was independently associated with a three‐fold higher chance of systolic nondipping BP underscores the potential impact of OSA in modulating cardiovascular risk in patients with hypertension regardless of sex.

5. Study Strengths and Limitations

Our study has some strengths and limitations that should be acknowledged. We used standard polysomnography, which is considered the gold standard method for diagnosing OSA. We also studied hypertensive patients who received 30 days of standardized antihypertensive treatment. However, even this careful standardization may not prevent the long‐term effects of previous treatment on arterial structure. Despite this limitation, previous antihypertensive treatments were similar in patients with and without OSA (Table S2). Instead of a single parameter of diastolic dysfunction commonly observed in several studies, we decided to use a more stringent criterion to define diastolic dysfunction. In addition, all measurements (including sleep studies) were performed in a blinded manner. The following limitations should be addressed. First, this study has a relatively small sample size. The reduced number of male patients without OSA may partially explain the lack of significant differences for PWV when we adjusted for 24‐hour and nighttime mean BP as well as for some cardiac parameters evaluated by echocardiography. The present findings apparently contradict those derived from our previous study that evaluated arterial stiffness and heart structure in men according to the presence of OSA and hypertension.8 In that study, we found that the presence of OSA and hypertension was associated with a higher percentage of left ventricular hypertrophy than observed in patients with hypertension and no OSA.8 However, only severe OSA was included, and diastolic function was not evaluated.8 Second, due to ethical reasons, all patients with hypertension were receiving medical treatment prior to entry into the study. In clinical practice, hypertensive patients who have never received treatment are difficult to identify, especially in a tertiary cardiology center. Third, these findings are applicable to patients with hypertension only. Finally, most of the women in the study were experiencing menopause. A recent investigation explored the impact of OSA in consecutive women in the perimenopausal period (75% of whom had hypertension).39 OSA was independently associated with high BP and increased arterial stiffness in these patients. This study, however, was limited to women in the perimenopausal period, did not perform echocardiographic analysis, did not compare data with men, and did not compare data in women with and without hypertension.

6. Conclusions

Our results reinforce the concept that the cardiovascular burden of OSA in patients with hypertension is not sex‐specific. Future studies will be necessary to compare the impact of OSA treatment with continuous positive airway pressure in both sexes with respect to several cardiovascular end points, including cardiovascular events and surrogate markers of cardiovascular risk.

Conflict of Interest

None.

Supporting information

 

 

 

Jenner R, Fatureto‐Borges F, Costa‐Hong V, et al. Association of obstructive sleep apnea with arterial stiffness and nondipping blood pressure in patients with hypertension. J Clin Hypertens. 2017;19:910–918. 10.1111/jch.13008