Sleep-disordered Breathing and Prevalent Hypertension in Men With and Without HIV

Abstract

Background:

Sleep-disordered breathing (SDB) is a known risk factor for hypertension. Despite the well-established link between HIV infection and hypertension, it remains to be determined whether HIV infection modifies the association between SDB and hypertension,.

Setting:

The Multicenter AIDS Cohort Study

Methods:

SDB was assessed using in-home polysomnography in 779 men (436 with and 343 without HIV). The apnea-hypopnea index (AHI) based on oxyhemoglobin desaturation threshold of ≥3% or arousal (AHI_3a) and ≥4% (AHI₄) along with oxygen desaturation index (ODI) were used to quantify SDB severity. Hypertension was defined as a blood pressure ≥140/90 mmHg, antihypertensive medication use, or self-report. The associations between HIV, SDB, and hypertension were characterized using multivariable logistic regression.

Results:

The prevalence of hypertension and SDB (AHI_3a≥5 events/hr) was high, with estimates of 53.8% and 82.8%, respectively. Among men without SDB, HIV was independently associated with hypertension, with an adjusted odds ratio (OR) of 3.05 (95%CI: 1.33–7.01). In men without HIV, SDB was associated with hypertension (OR: 2.93; 95%CI: 1.46–5.86). No significant increase in the odds of hypertension was noted in men with both HIV and SDB compared to men with either factor alone, with an OR of 3.24 (95%CI: 1.62–6.47). These results were consistent across different measures used to define SDB (AHI_3a, AHI₄, ODI₃, and ODI₄).

Conclusion:

Predictors of hypertension differed by HIV status. SDB was associated with hypertension in men without HIV, but not in men with HIV. Among men with HIV, SDB did not affect the odds of hypertension.

INTRODUCTION

Sleep-disordered breathing (SDB) is a prevalent condition that affects approximately 9% to 38% of adults in the general population^1,2. It is well established that SDB is more common in men than women and is associated with being overweight or obese.^3–6 Hypertension, another widely prevalent condition, exhibits similar demographic patterns to those of SDB.⁷ Risk factors common to SDB and hypertension include male sex, obesity, and older age. Approximately 30%−50% of patients with hypertension have SDB and nearly 50% of patients with SDB have hypertension.^4,8–10 Treatment of SDB with positive airway pressure (PAP) therapy modestly improves both systolic and diastolic blood pressure.¹¹ Meta-analyses of randomized clinical trial data show that the effects of PAP therapy on blood pressure are most notable in patients with severe hypertension, with an average decrease of 6–7 mmHg and 4–5 mmHg in systolic and diastolic blood pressure, respectively.^11,12

SDB and hypertension are also common in people living with human immunodeficiency virus (HIV). In the Multicenter AIDS Cohort Study (MACS), SDB was noted in 58.6% of men with HIV and was more common than in men without HIV.¹³ Hypertension affects approximately 25% of people living with HIV¹⁴, with age and the use of antiretroviral therapy as independent risk factors.^15–19 In fact, over 50% of people with HIV receiving antiretroviral therapy and over the age of 50 years have hypertension.²⁰ Despite the high burden of SDB and hypertension in people with HIV, studies to date have not assessed whether SDB and HIV synergistically influence the prevalence of hypertension. Thus, the overarching objective of this study was to investigate the independent and interactive associations between HIV, SDB, and hypertension in a cohort of men with and without HIV. The study also sought to characterize the associations between the degree of SDB-related nocturnal hypoxemia, frequency of arousals, and hypertension in men with and without HIV.

METHODS

Study Population

The MACS is a longitudinal cohort study of the natural and treated history of HIV disease in men who have sex with men, now part of the MACS-WIHS Combined Cohort Study (MWCCS). MACS participants were recruited from Baltimore/Washington DC, Chicago, Pittsburgh/Columbus, and Los Angeles, as previously reported.²¹ The MACS is comprised of men both living with and without HIV, and was conducted spanning four waves of enrollment (1984–1985, 1987–1991, 2001–2003, and 2010–2017). Semiannual study visits were conducted using standardized interviews which included assessment of health behaviors (e.g., smoking, alcohol use), physical examination and anthropometry, and measurement of T-cell subsets and plasma HIV RNA concentration. Blood pressure was measured during the study visit by trained clinicians as previously described.²² Use of medications (e.g., antihypertensives) was also recorded at each visit, as was self-reported diagnoses of conditions such as hypertension. Between March 2018 and June 2019, a nested study was conducted within the MACS to evaluate for SDB using home polysomnography.^13,23 Participants who were enrolled in the MACS were eligible for inclusion and those who expressed interest in participating were asked to provide informed consent. The study protocol was approved by an Institutional Review Board at each of the participating sites.

Home Polysomnography

The protocol for home polysomnography and participant instruction have been previously described.²³ Briefly, the home polysomnogram was conducted using a self-applied recorder (Nox A1, Nox Medical, Reykjavik, Iceland), which collected a frontal electroencephalogram (EEG) montage with the following derivations: AF4, AF3, AF7, and AF8. Additional physiological data recorded included the frontalis muscle electromyogram (EMG), the electrocardiogram (ECG), right and left anterior tibialis electromyogram (EMG), nasal airflow using a pressure transducer, pulse oximetry, and chest and abdominal respiratory movement. The digital data from the Nox A1 recorder were transmitted to a central reading facility for manual scoring of sleep and disordered breathing events. Sleep was classified as wake, non-rapid eye movement (stage N1, N2, and N3) sleep, or rapid eye movement (REM) sleep using 30-second epochs. Apneas were identified if there was an absence or near absence of airflow for at least ten seconds. Hypopneas were identified as a reduction in airflow by 30% or more for at least 10 seconds, along with a decrease of ≥3% in oxygen saturation. An alternative definition of hypopnea was also used, requiring a 4% decrease in oxygen saturation or an arousal from sleep as detected in the EEG. The apnea-hypopnea index (AHI) was calculated as the number of apneas and hypopneas per hour detected during sleep. The oxygen desaturation index (ODI) was calculated as the number of ≥ 3% desaturations per hour detected during sleep. A more stringent definition of ODI that required a desaturation of ≥4% was also used. Thus, four distinct measures were used to define SDB: (a) ODI based on the 3% oxygen desaturation threshold (ODI₃); (b) ODI based on the 4% oxygen desaturation threshold (ODI₄); (c) AHI based on the hypopnea definition of 3% oxygen desaturation or an EEG arousal (AHI_3a); and (d) AHI based on the hypopnea definition using the 4% oxygen desaturation threshold (AHI₄). For each of these measures, SDB was considered present if the AHI was ≥ 5 events/hr. Severity of SDB was categorized as mild (5.0–14.9 events/hr), moderate (15.0–29.9 events/hr), or severe (≥ 30 events/hr). To evaluate the degree of nocturnal hypoxemia, the total sleep time with an oxygen saturation below 90% (TST_90%) was determined. SDB-related sleep fragmentation was assessed with the frequency of arousals per hour of sleep.

Statistical Analyses

Prevalent hypertension was defined as an elevated blood pressure (systolic ≥ 140 mm Hg or diastolic ≥ 90 mm Hg), use of antihypertensive medications, or a self-reported clinical diagnosis based on data from the visit closest to the sleep assessment. Hypertension prevalence was determined among men with and without SDB stratified by HIV status. The association between hypertension and SDB was determined using each of the four SDB measures (i.e., AHI_3a, AHI₄, ODI₃, and ODI₄). To account for demographic and anthropometric differences between men with and without HIV, adjusted prevalence estimates were derived using logistic regression models that included age, race, and BMI as covariates. Race was included as a categorical variable (Black, White, and Other) with White race as the reference. Multivariable logistic regression models were also used to investigate the associations between HIV status, SDB severity, and hypertension. Effect modification between HIV status and SDB severity (none, mild, moderate, or severe) was examined by including interaction terms between the two variables in the multivariable models. Similarly, interactions terms between SDB and antiretroviral therapy class (i.e., integrase inhibitor, non-nucleoside reverse transcriptase inhibitor, or protease inhibitor) were also considered for models that were restricted to men with HIV. The resulting adjusted odds ratios and 95% confidence intervals (95% CI) from these models are reported. To delineate the independent associations between SDB-related nocturnal hypoxemia, sleep fragmentation, and hypertension, TST_90% and the arousal frequency were included, respectively, as the primary independent variables in separate multivariable logistic regression models along with their corresponding interaction terms with HIV status. All analyses were conducted using Stata 17.0 software (Stata Inc., College Station TX) and statistical significance was set at an α value of 0.05.

RESULTS

The study sample included 851 men who completed home polysomnography. After excluding 55 men who had missing BMI data and 17 men with missing data on hypertension, complete information on demographics, anthropometry, and home polysomnography were available for 779 men. In that subset, 436 (56.0%) were men with HIV and 343 (44.0%) men without HIV. Compared to men without HIV, men with HIV had a similar BMI, but were younger and more likely to be current smokers and nonwhite (Table 1). The vast majority of men with HIV were receiving antiretroviral therapy (92.6%) and were virologically suppressed (92.4%) with a median CD4 cell count of 700/μl. In men with HIV, 46.6% were receiving integrase inhibitors, 24.8% were receiving non-nucleoside reverse transcriptase inhibitors, and 20.6% were receiving protease inhibitors. These characteristics were comparable to those of the MACS population as a whole (data not shown). The unadjusted prevalence of hypertension was 53.8% (52.1% in men with HIV and 56.0% in men without HIV; p=0.28). SDB prevalence was also high with unadjusted estimates of 82.8%, 53.4%, 77.3%, and 51.0% based on the AHI_3a, AHI₄, ODI₃, and ODI₄ values being ≥ 5 events/hr, respectively. The unadjusted prevalence of SDB did not differ significantly between men with and without HIV for any of the four measures of SDB.

Table 1:

Characteristics on men with and without HIV in the Multi-Center AIDS Cohort Study.

	Men with HIV (N=436)	Men without HIV (N=343)	p-value†
Age*, years	55.5 (49.0–63.0)	62.0 (56.0–68.0)	< 0.001
Body mass index*, kg/m2	27.0 (24.1–30.5)	26.6 (23.6–30.3)	0.18
Race, %
White	56.0%	77.2%	< 0.001
Black	32.8%	17.8%
Other	11.2%	5.0%
Smoking status
Never	34.0%	35.0%	0.001
Former	43.5%	52.7%
Current	22.5%	12.4%
HIV viral load <200 copies/ml	92.4%	-
Antiretroviral therapy	92.6%	-
Integrase inhibitor	46.6%	-
NNRTI‡	24.8%	-
Protease inhibitor	20.6%	-
CD4 cell count (/μl)	700 (517–900)	-
HIV viral load (copies/ml)	<20

^*Values reported are medians (interquartile range).

^†p-value for comparing men with and without HIV

^‡NNRTI: Non-nucleoside reverse transcriptase inhibitor.

To assess the independent and interactive associations between HIV, SDB, and hypertension, multivariable logistic regression models were used with HIV and SDB as independent predictors along with an interaction term between the two (Table 2). Compared to men with neither HIV or SDB (reference category), men without HIV but with SDB (AHI_3a ≥ 5 events/hr) were significantly more likely to have hypertension, with an adjusted odds ratio of 2.93 (95% CI: 1.46–5.86). Men with HIV, but without SDB (AHI_3a < 5 events/hr), were also significantly more likely to have hypertension, with an adjusted odds ratio of 3.05 (95% CI: 1.33–7.01). Thus, both SDB and HIV were independently associated with hypertension. In men with both HIV and SDB, the adjusted odds ratio for hypertension was 3.24 (95% CI: 1.62–6.47), indicating that the combination of HIV and SDB was not associated with a statistically significant higher prevalence of hypertension than that seen with either factor alone. The negative interaction between SDB and HIV on the odds of hypertension was statistically significant or borderline significant for all four measures of SDB (Supplement Table S1), including AHI_3a (p=0.024), AHI₄ (p=0.022), ODI₃ (p=0.013), and ODI₄ (p=0.065),

Table 2:

Adjusted^* odds ratios (95% CI) for hypertension according to HIV and SDB status.

HIV status†	SDB status‡	Odds Ratio	95% CI
−	−	Reference
−	+	2.93	1.46–5.86
+	−	3.05	1.33–7.01
+	+	3.24	1.62–6.47

^*Adjusted odd ratio from multivariable logistic regression model with age, race, and BMI.

^†Men with HIV (+) and men without HIV (−) .

^‡AHI_3a ≥ 5 events/hr (SDB status +) and AHI_3a < 5 events/hr (SDB status −)

Additional analyses were undertaken to characterize the effects of HIV and SDB on hypertension by including severity of SDB (none, mild, moderate, and severe) in the multivariable model. In men without HIV (Figure 1A, circles), SDB severity was associated with hypertension independent of age, race, and BMI. Specifically, the adjusted prevalence of hypertension was lowest (32.0%) in those without SDB (i.e., AHI_3a < 5 events/hr), intermediate in those with mild (52.6%) and moderate SDB (48.9%), and highest (65.9%) in those with severe SDB (AHI_3a ≥ 30 events/hr). In men with HIV (Figure 1A, squares), the adjusted prevalence of hypertension in those without SDB AHI_3a < 5 events/hr) was 54.6%. This was significantly higher than the prevalence in men with neither HIV nor SDB (32.0%; p=0.005) and corresponded to an adjusted odds ratio for the association between HIV and prevalent hypertension of 3.05 (95% CI:1.33–7.01). Hypertension prevalence in men with HIV was similar in all categories of SDB severity (Figure 1A, squares). The associations between HIV, SDB, and hypertension were essentially unchanged when other measures of SDB (i.e., AHI₄, ODI₃, and ODI₄) were examined as shown in Figures 1B, 1C, and 1D, respectively. Finally, in analyses restricted to men with HIV, the use or the type of antiretroviral therapy (i.e., integrase inhibitor, non-nucleoside reverse transcriptase inhibitor, or protease inhibitor) was not associated with hypertension, either independently or in an interaction with SDB severity (data not shown).

Figure 1:

Adjusted prevalence and 95% confidence intervals for hypertension by HIV status and severity of sleep-disordered breathing (SDB) as a categorical variable (<5.0, 5.0–14.9, 15.0–29.9, and ≥30.0 events/hr) for AHI_3a (A), AHI₄ (B), ODI₃ (C), and ODI₄ (D). Prevalence estimates were adjusted for age, race, and BMI. Circles represent men without HIV and squares represent men with HIV.

Men with and without HIV had similar severity of nocturnal hypoxemia (TST₉₀) with a median TST₉₀ of 1.6% (interquartile range (IQR) 0.1–9.1%) in men with HIV and 1.4% (IQR 0.2–10.0%) in men without HIV. The corresponding values for the arousal index were 13.4 events/hr (IQR: 9.5–19.4 events/hr) and 13.2 events/hr (IQR: 9.3–18.5 events/hr). After adjusting for age, BMI, and race in multivariable logistic models, TST₉₀ was a significant predictor of hypertension in men without HIV, but not in men with HIV, indicating that the association between TST₉₀ and hypertension was dependent on HIV status (Table 3). The arousal frequency showed no significant association with hypertension in either group of men, whether it was considered as a categorical variable (Table 4) or a continuous variable (data not shown).

Table 3:

Adjusted^* odds ratios for hypertension for quartiles of TST₉₀ by HIV status.

Group	Quartile of TST90 (%)
	I (≤ 0.1%)	II (0.2% – 1.5%)	III (1.6% – 9.7%)	IV (> 9.7%)
Men without HIV	Reference	1.21 (0.64 – 2.29)	1.37 (0.69 – 2.71)	2.04 (1.00 – 4.19)
Men with HIV	Reference	0.87 (0.47 – 1.58)	0.82 (0.46 – 1.47)	1.16 (0.62 – 2.16)

^*Adjusted for age, race, and BMI using multivariable logistic regression models for prevalent hypertension.

Table 4:

Adjusted^* odds ratios for hypertension for quartiles arousal frequency quartiles by HIV status.

Group	Quartile of Arousal Frequency (events/hr)
	I (< 9.4 events/hr)	II (9.4–13.3 events/hr)	III (13.4–19.1 events/hr)	IV (> 19.1 events/hr)
Men without HIV	Reference	1.06 (0.56 – 2.00)	1.35 (0.72 – 2.56)	1.54 (0.80 – 2.99)
Men with HIV	Reference	1.30 (0.71 – 2.38)	1.12 (0.62 – 2.04)	0.99 (0.55 – 1.77)

^*Adjusted for age, race, and BMI using multivariable logistic regression models for prevalent hypertension.

DISCUSSION

The results of this study on the association between SDB and hypertension in men with and without HIV show that after adjustments for age, BMI, and race, both HIV and SDB were independently associated with hypertension. Among men without HIV, SDB was associated with prevalent hypertension. While HIV was also found to be associated with hypertension, the current study did not reveal an additive effect of SDB on this association. The odds of hypertension were found to be similar among men with both HIV and SDB as compared to those with only one of these factors. These findings were similar regardless of the measure used to define SDB (i.e., AHI_3a, AHI₄, ODI₃, or ODI₄,). Of the pathophysiological consequences of SDB, severity of nocturnal hypoxemia (TST₉₀) was found to be associated with hypertension in men without HIV, but not in men with HIV. In contrast, no significant association was noted between the frequency of arousals and hypertension in the overall sample or in the subgroups of men with and without HIV.

The epidemiology of hypertension in people with HIV has yielded conflicting results.^18,19 Although studies conducted in North America have reported an association between HIV and a higher prevalence of hypertension, studies conducted in other parts of the world have reported contradictory results (i.e., lower prevalence of hypertension).¹⁸ The discrepancy in the observed association between HIV and hypertension across studies is likely due to differences in factors such as age, use of antiretroviral therapy, lifestyle, and access to health care. In contrast, studies conducted in the general population over the last two decades have consistently found a link between SDB and prevalent and incident hypertension.²⁴ Moreover, a meta-analysis comprising 44 randomized clinical trials has demonstrated that treatment of SDB with PAP therapy can lower systolic blood pressure by 2.1 mm Hg (95% CI: 1.4–2.8 mm Hg) and diastolic blood pressure by 1.9 mm Hg (95% CI: 1.4–2.40 mm Hg).²⁵ In patients with refractory hypertension, SDB is highly prevalent and treatment with PAP therapy has been shown to significantly lower blood pressure, indicating that the poor response to antihypertensive medications is related to untreated SDB.²⁶ Thus, there is a strong empirical evidence base supporting the notion that SDB leads to the development of hypertension in the general population. SDB increases this risk through several mechanisms.²⁷ Nocturnal hypoxemia and swings in intrathoracic pressure in SDB can lead to chemoreceptor activation and increase in sympathetic nervous system activity^27,28. Intermittent hypoxemia due to SDB can also increase oxidative stress, systemic inflammation, and metabolic dysfunction, resulting in endothelial dysfunction and vascular remodeling^27,29. Additionally, SDB-related intermittent hypoxemia can activate the renin-angiotensin-aldosterone system (RAAS) and increase aldosterone production, which can also contribute to the development of hypertension^27,29.

Several of the aforementioned mechanisms may also be relevant to people with HIV. For example, individuals with HIV may experience chronic immune activation, as evidenced by elevated levels of interleukin-6³⁰, sCD14, and sCD163³¹, all of which have been linked to hypertension. Increased activity of RAAS has also been observed in people with HIV.^32,33 Additionally, hypertension in this population may be influenced by HIV infection itself and the use of antiretroviral therapy, particularly protease inhibitors. These agents are known to activate the RAAS³² and cause lipodystrophy^34,35. Protease inhibitors can also lead to endothelial dysfunction, arterial stiffness, and dyslipidemia thereby contributing to the higher the risk of hypertension.^36–38 Certain nucleoside reverse transcriptase inhibitors may also augment the risk of hypertension, as can immune reconstitution resulting from antiretroviral therapy.³⁷ Despite the evidence linking antiretroviral therapy to hypertension, the current study did not find an association between its use and hypertension. This could be due to the fact that only a small proportion of the participants (7.4%) of men with HIV were not on antiretroviral therapy.

The current study has several notable strengths. First, the use of full montage home polysomnography allowed for a rigorous assessment of SDB prevalence, severity, and related factors, including degree of nocturnal hypoxemia and sleep fragmentation, in a large sample of men with and without HIV. Second, the study population was well-characterized in terms of demographic and anthropometric data, providing a unique opportunity to characterize the independent associations among SDB, HIV, and hypertension. Third, the inclusion of demographically similar men with and without HIV in the MACS offered an optimal comparison group, minimizing the effects of confounding factors on the association between HIV and SDB. Finally, given that the MACS-WIHS Combined Cohort Study involves ongoing follow-up visits, analyses of incident hypertension in relation to SDB and HIV status will be possible in the future. This study also has several limitations that warrant discussion. Because the MACS included only men, the generalizability of the findings is limited. Additionally, the majority of the men in the study were virologically suppressed, which differs from people with HIV who are not receiving effective antiretroviral therapy. Furthermore, cohort enrollment bias and survivor bias may affect the representativeness of the MACS population as it pertains to contemporary men who have sex with men. An additional limitation is that the high prevalence of SDB in men with and without HIV could have obscured the identification of an additive effect of HIV and SDB on prevalent hypertension. Lastly, the use of older antiretroviral agents in some of the study participants may not reflect current HIV treatment practices.

The above limitations notwithstanding, the current study shows independent associations between HIV, SDB, and hypertension, with a negative interaction between SDB and HIV such that the association between SDB and hypertension was present only in men without HIV. The reasons for this unexpected finding are unclear, but living with HIV may saturate the mechanisms linking SDB to hypertension. Nevertheless, given the high prevalence of SDB in men with or at risk for HIV, identifying and treating SDB in this population is clinically relevant due to the established benefits of SDB treatment on improving blood pressure, daytime sleepiness, and quality of life in those without HIV.^39,40 It remains to be determined if SDB in HIV is also linked to incident hypertension and other cardiovascular outcomes, such as myocardial infarction, heart failure, and stroke. Additional research is clearly needed to address this knowledge gap as the medical comorbidity associated with HIV is now being increasingly recognized.³⁹