Associations between sleep apnea and subclinical carotid atherosclerosis: The Multi-Ethnic Study of Atherosclerosis

Abstract

Background and Purpose

Many health effects of sleep apnea (SA) may be mediated through accelerated atherosclerosis. We examined the associations of snoring and several measurements of SA with subclinical carotid atherosclerosis in a large multiethnic population sample.

Methods

This analysis included 1615 participants (mean age 68 years) from examination 5 (2010-2013) of the Multi-Ethnic Study of Atherosclerosis. Sleep measures including SA (apnea-hypopnea index [AHI 4%] ≥15 events/hour) were derived from full in-home polysomnography. Carotid atherosclerosis was measured using high-resolution B-mode ultrasound. Multivariable linear and logistic regression models were used to evaluate the associations between sleep exposures with carotid intima-media thickness (CIMT) and the presence of carotid plaque, respectively. Effect modification by age, sex, and race/ethnicity was examined.

Results

In multivariable analysis, SA was associated with an increased odds of carotid plaque presence in individuals aged <68 years (odds ratio [OR], 1.47; 95% confidence interval [CI] 1.05 to 2.06) but not in older individuals (OR 0.95; 95% CI 0.67 to 1.37, p-interaction=0.078). Greater hypoxemia (sleep time <90% saturation) was associated with increasing CIMT in younger (0.028 ± 0.014 mm) but not in older individuals (−0.001 ± 0.013 mm, p-interaction= 0.106). Self-reported snoring was not associated with carotid atherosclerosis. In assessing race-specific outcomes, greater hypoxemia was associated with increased CIMT in Blacks (0.049 ±0.017 mm, p-interaction=0.033).

Conclusions

In this large multiethnic population-based sample, sleep disturbances are associated with subclinical carotid atherosclerosis in both men and women, particularly in those <68 years. The mechanisms underlying the association between sleep apnea and carotid atherosclerosis may differ for carotid plaque and CIMT.

Introduction

Stroke is a leading cause of disability, mortality, and economic burden in the United States.¹ Obstructive sleep apnea (OSA), a condition characterized by repetitive upper airway closure during sleep, affects an estimated 24% of men and 9% of women in the general population and may affect up to 50-70% of stroke patients.^1–3 There is a strong association between OSA and increased risk of stroke,^4–9 however, associations may be stronger in men than women.⁴ The presence of OSA is also associated with higher mortality and worse functional outcomes following stroke.¹⁰

Carotid intima-media thickness (CIMT) and carotid plaque are widely accepted non-invasive measures of atherosclerosis and subclinical arterial injury. Prior studies have demonstrated that CIMT and carotid plaque presence or burden are associated with increased incidence of cardiovascular events, including stroke, in individuals without prior history of cardiovascular disease (CVD).^11–17

Several studies have demonstrated an association between SA and increased CIMT.¹⁸ However, most of these previous studies were small and did not fully control for potential confounders and did not study an ethnically diverse sample. Furthermore, few studies have evaluated the relationship between SA and carotid plaque, which more recent studies have found to be a stronger predictor of incident CVD than CIMT.¹⁹

The objective of this study was to examine the association between snoring, SA and subclinical carotid atherosclerosis in a large multiethnic population sample and assess variations by age, sex, and race/ethnicity. Given stronger associations between CVD and SA in middle-aged compared to older individuals^20–22 and in men compared to women,⁴ we hypothesized that the associations between subclinical atherosclerosis with SA would be stronger in younger individuals and in men, as compared to their counterparts.

Methods

Anonymized data and materials have been made publicly available at BioLINCC and can be accessed at https://biolincc.nhlbi.nih.gov/home/. Polysomnography data are available at the National Sleep Research Resource and can be accessed at https://sleepdata.org.

See Supplemental Material for detailed methods.

Study Population

The Multi-Ethnic Study of Atherosclerosis (MESA) is a prospective study designed to investigate risk factors for incident CVD and progression of subclinical CVD in an ethnically diverse population.²³ Between July 2000 and August 2002, 6,814 men and women aged 45-85 and free of clinically apparent CVD were recruited from six U.S. centers. At MESA Exam 5 (2010 to 2013), eligible participants were invited to participate in the MESA Sleep Study. The present cohort consists of 1,615 participants with successful polysomnography (PSG) and carotid ultrasound data at Exam 5. The study was approved by the Institutional Review Board at each center and written informed consent was obtained from all participants.

Polysomnography

Detailed MESA PSG methodology has been published.²⁴ Sleep stages and arousals were scored by the Brigham and Women’s Hospital Sleep Reading Center according to published guidelines²⁵ adapted from the Sleep Heart Health Study^26,27. Apneas were scored when the thermocouple signal flattened or nearly flattened for greater than 10 seconds. Hypopneas were scored if there was a >30% reduction in amplitude of either respiratory effort or airflow for ≥10 seconds. Sleep apnea was defined by an apnea-hypopnea index (AHI) ≥15 events/h (all apneas and hypopneas associated with ≥4% oxygen desaturation).

Carotid Ultrasonography

B-mode ultrasound images of the carotid artery were reviewed and interpreted by the University of Wisconsin Atherosclerosis Imaging Research Program MESA Carotid Ultrasound Reading Center.²⁸ The distal common carotid artery (CCA) was defined as the distal 10 mm of the vessel. IMT was defined as the IMT measured as the mean of the mean left and right mean far wall distal CCA wall thicknesses. Carotid plaque score (0-12) was defined as the number of carotid plaques in the internal, bifurcation, and common segments of both carotid arteries. Carotid plaque was defined as a discrete, focal wall thickening ≥1.5 cm or focal thickening >50% greater than the surrounding IMT.²⁸

Other Measures

Habitual snoring was considered if participants reported snoring sometimes (3-5 nights/week) or always/almost always (6-7 nights/week) on the sleep questionnaires.

Covariates

Race/ethnicity was self-identified as White, Black, Hispanic, and Chinese.

Statistical Analyses

Data are reported as mean (SD) for continuous variables and percentages for categorical variables. Between group comparisons were made using Chi-square tests, t-tests, and Wilcoxon rank-sum tests for categorical, normally and non-normally distributed continuous variables, respectively.

Linearity of dose-response relationship between AHI and mean CCA was assessed by fitting a linear regression model using quartiles of AHI as a categorical variable and assessing linearity by graphical examination and by likelihood ratio testing.

Logistic regression models were used to assess the association between sleep apnea (AHI ≥15 events/hour) and other sleep measures with the presence of carotid plaque, adjusting for age, race/ethnicity, and sex (model 1); further adjusting for body mass index (BMI) and pack-years smoked (model 2), and additionally adjusting for factors that could be confounders or potentially in the causal pathway including alcohol use, the presence of hypertension and diabetes, serum total cholesterol and triglyceride level, and use of statin medications (model 3). Similarly, linear regression models were used to examine the association between sleep apnea and other sleep measures with mean CIMT. We also examined the relationship between sleep time with oxygen saturation <90% (SpO₂), a non-normally distributed variable, and measures of carotid atherosclerosis, by analyzing sleep time with SpO₂ <90% as a dichotomized variable (median of ≥0.64%). Since age, sex, race/ethnicity may be important effect modifiers of the relationship between AHI and carotid atherosclerosis, we assessed if the associations varied by age, sex or race/ethnicity by adding interaction terms into the models with age dichotomized at the median of 68 years. Statistical analysis was performed using SAS version 9.4 (Cary, NC).

Results

Baseline characteristics of the sample are shown in Table 1. The mean age (SD) of the cohort was 68.7 (9.1) years, 54.1% were women and 64.5% were non-whites. Moderate to severe SA (AHI ≥15) was present in 32.9% of participants. Participants with SA were more likely men, had higher BMI, and greater CVD risk factors including hypertension and diabetes, than those without SA. The mean (SD) CIMT was 0.86 mm (0.20) and 66.4% of participants had one or more carotid plaques with a median carotid plaque score of 1 (IQR 0, 4). Individuals with SA were more likely to have carotid plaque and higher CMIT than those without SA.

Table 1.

Baseline Characteristics by the Presence of Sleep Apna (SA: AHI ≥15 events/hr)

	Total Sample N=1615	SA (AHI ≥15) N=531	No SA (AHI <15) N=1084	P value
Age (years)    Mean ± SD    Median [IQR]	68.7 ± 9.1 68 [61, 76]	69.1 ± 8.9 69.0 [62, 76]	68.4 ± 9.2 68.0 [61, 75]	0.108
Male, n (%)	741 (45.9)	325 (61.2)	416 (38.4)	<0.0001
Race, n (%)    White    Chinese    Black    Hispanic	573 (35.5) 207 (12.8) 447 (27.7) 388 (24.0)	170 (32.0) 81 (15.3) 131 (24.7) 149 (28.1)	403 (37.2) 126 (11.6) 316 (29.2) 239 (22.1)	0.003
BMI, kg/m2	28.6 (5.5)	30.5 ± 5.7	27.7 ± 5.1	<0.0001
Smoking status, n (%)    Never    Former    Current	759 (47.2) 740 (46.0) 110 (6.8)	245 (46.3) 255 (48.2) 29 (5.5)	514 (47.6) 485 (44.9) 81 (7.5)	0.215
Pack years smoked	0 (0, 12.4)	0 (0, 12.5)	0 (0, 12.2)	0.620
Current alcohol use, n (%)	690 (42.9)	232 (43.9)	458 (42.4)	0.581
Lipid lowering medication use, n (%)	612 (37.9)	221 (41.6)	391 (36.1)	0.031
Antihypertensive use, n (%)	873 (54.1)	314 (59.1)	559 (51.6)	0.004
Presence of hypertension, n (%)	973 (60.3)	344 (64.8)	629 (58.0)	0.009
Mean systolic blood pressure, mmHg	123.1 ± 20.3	124.2 ± 19.4	122.5 ± 20.7	0.063
Mean diastolic blood pressure, mmHg	68.1 ± 9.8	69.3 ± 9.6	67.6 ± 9.8	0.002
Presence of diabetes, n (%)	352 (22.1)	148 (28.1)	204 (19.1)	<0.0001
Serum total cholesterol, mg/dL	183.9 ± 36.2	179.7 ± 36.9	186.0 (35.7)	0.003
Serum low-density lipoprotein, mg/dL	106.4 ± 32.0	104.4 ± 32.8	107.4 ± 31.6	0.156
Serum high-density lipoprotein, mg/dL	55.8 ± 16.4	51.5 ± 14.0	57.9 ± 17.2	<0.0001
Serum triglyceride, mg/dL	110.0 ± 61.8	120.5 ± 69.3	104.8 ± 57.1	<0.0001
HMG-CoA Reductase (statin) use, n (%)	581 (36.0)	211 (39.7)	370 (34.1)	0.028
Self-reported snoring, n (%)    None habitual snorer (1-2 nights/week)    Habitual snorer (3-7 nights/week)    Unknown	449 (28.0) 644 (40.1) 512 (31.9)	105 (19.8) 270 (50.9) 155 (29.3)	344 (32.0) 374 (34.8) 357 (33.2)	<0.0001
Carotid ultrasound measures
Total carotid plaque score, median [IQR]	1 [0, 4]	2.0 [0, 4]	1.0 [0, 3]	0.015
Presence of carotid plaque, n (%)	1073 (66.4)	373 (70.2)	700 (64.6)	0.023
Mean common carotid intima-media thickness, mm       Mean ± SD       Median [IQR]	0.86 ± 0.20 0.82 [0.73, 0.96]	0.88 ± 0.19 0.84 [0.74, 0.99]	0.85 ± 0.20 0.82 [0.72, 0.95]	0.001
Polysomnography sleep measures
Total sleep duration, min	360.8 ± 81.8	350.0 ± 83.4	366.3 ± 80.5	0.0001
Sleep maintenance efficiency, %	79.7 ± 12.6	76.6 ± 13.3	81.3 ± 11.9	<0.0001
Arousal index, events/h	22.0 ± 11.9	29.6 ± 13.9	18.3 ± 8.7	<0.0001
Sleep time in stage 1 sleep, %	14.2 ± 8.9	19.0 ± 11.0	11.9 ± 6.5	<0.0001
Sleep time in stage 2 sleep, %	57.6 ± 10.1	56.4 ± 10.6	58.2 ± 9.7	0.008
Sleep time in slow wave sleep (N3), %	10.2 ± 9.0	8.3 ± 8.0	11.1 ± 9.2	<0.0001
Sleep time in REM sleep, %	18.1 ± 6.7	16.3 ± 7.1	18.9 ± 6.3	<0.0001
Sleep time with SpO2 <90%, %, median	0.64 [0.1, 3.1]	4.0 [1.6, 9.4]	0.2 [0, 1.0]	<0.0001
Average SpO2 in sleep, %, median [IQR]	94.8 (93.5, 95.7)	93.8 (92.7, 94.8)	95.0 (94.0, 96.0)	<0.0001

Data are presented as number (%), mean ± SD, or median [IQR].

P values for continuous data are from Wilcoxon rank sum test. P values for categorical data are from a chi-square test.

AHI indicates apnea hypopnea index; BMI, body mass index; IQR, interquartile range; REM, rapid eye movement sleep; SD, standard deviation; SA, sleep apnea; and SpO₂, oxygen saturation.

Participant characteristics by the presence carotid plaque are shown in Supplemental Table I. Individuals with carotid plaque had higher AHI levels, more sleep hypoxemia, and more fragmented sleep as noted by lower sleep maintenance efficiency, higher arousal index and more N1 and less REM sleep.

Carotid Plaques

After adjusting for demographic factors, a positive but nonsignificant association between SA and carotid plaque was observed, which was attenuated after adjustment for BMI and smoking with further attenuation after adjusting for multiple cardiovascular risk factors and medications (Table 2). A similar trend was observed for the association between arousal index and carotid plaque. In adjusted analyses, there was no evidence for significant associations between hypoxemia, sleep efficiency, sleep duration, or slow wave sleep duration in the overall cohort.

Table 2.

Association between Sleep Apnea and Other Sleep Measures and the Presence of Carotid Plaque

	Model 1 OR [95% CI]	P-value	Model 2 OR [95% CI]	P-value	Model 3 OR [95% CI]	P-value
SA (AHI ≥15 events/h)	1.26 [0.99, 1.60]	0.056	1.22 [0.94, 1.56]	0.130	1.16 [0.90, 1.51]	0.250
Arousal index (events/h)	1.01 [1.00, 1.02]	0.076	1.01 [1.00, 1.02]	0.132	1.01 [1.00, 1.02]	0.143
Sleep maintenance efficiency, %	1.00 [0.99, 1.00]	0.329	1.00 [0.99, 1.01]	0.402	1.00 [0.99, 1.01]	0.497
≥0.64% sleep time with SpO2 <90%, %	1.13 [0.91, 1.41]	0.254	1.04 [0.82, 1.32]	0.723	1.02 [0.80, 1.30]	0.873
Total sleep duration, min	1.00 [1.00, 1.00]	0.521	1.00 [1.00, 1.00]	0.671	1.00 [1.00, 1.00]	0.768
Slow wave sleep (N3), %	1.00 [0.99, 1.02]	0.686	1.00 [0.99, 1.02]	0.606	1.00 [0.99, 1.02]	0.579

^†Model 1: adjusted for demographic factors (continuous age, race/ethnicity, sex).

Model 2: additionally adjusted for pack-years smoked and body mass index

Model 3: additionally adjusted for alcohol use, presence of hypertension and diabetes mellitus, serum total cholesterol, serum triglyceride level, HMG-CoA reductase (statin) use.

AHI indicates apnea hypopnea index; OR, odds ratio; SA, sleep apnea; and SpO₂, oxygen saturation.

Analysis that included the potential for moderation by age and adjusted for demographics, obesity and smoking showed that SA was associated with a 51% increased odds of carotid plaque in individuals aged <68 years (OR 1.51; 95% CI 1.08 to 2.10) but not in older individuals (OR 1.02; 95% CI 0.72 to 1.46; p-interaction=0.105). The association persisted in the younger group after further adjusting for major cardiovascular risk factors (OR 1.47; 95% CI 1.05 to 2.06), with no evidence of association in the older sample (p-interaction=0.078) (Table 3). Although the test for age interaction was not significant, there also appeared to be a positive association between increasing arousal index and carotid plaque in individuals age <68 years, which was not observed in older individuals. Stronger associations between hypoxemia and carotid plaque were suggested in younger than older individuals in analyses adjusted for demographic factors; however, these associations were not significant in more fully adjusted models.

Table 3.

Association between Sleep Apnea and Other Sleep Measures and the Presence of Carotid Plaque by Dichotomized Age

	Model 1 OR [95% CI] P-value	P-interaction	Model 2 OR [95% CI] P-value	P-interaction	Model 3 OR [95% CI] P-value	P-interaction
SA (AHI ≥15 events/h)
Age <68	1.56 [1.14, 2.13] 0.006	0.064	1.51 [1.08, 2.10] 0.015	0.105	1.47 [1.05, 2.06] 0.027	0.078
Age ≥68	1.01 [0.71, 1.42] 0.968		1.02 [0.72, 1.46] 0.892		0.95 [0.67, 1.37] 0.798
Arousal index (events/h)
Age <68	1.02 [1.00, 1.03] 0.020	0.252	1.02 [1.00 1.03] 0.020	0.404	1.01 [1.00, 1.03] 0.035	0.264
Age ≥68	1.00 [0.99, 1.02] 0.577		1.00 [0.99, 1.02] 0.577		1.00 [0.99, 1.02] 0.664
≥0.64% Sleep time with SpO2 <90%, %
Age <68	1.37 [1.03, 1.82] 0.033	0.086	1.29 [0.95, 1.75] 0.108	0.119	1.25 [0.91, 1.71] 0.166	0.127
Age ≥68	0.94 [0.68, 1.29] 0.688		0.91 [0.65, 1.27] 0.568		0.88 [0.63, 1.24] 0.463

^†Model 1: adjusted for demographic factors (dichotomized age, race/ethnicity, sex) and interaction term between age and sleep parameter.

Model 2: additionally adjusted for pack-years smoked and body mass index

Model 3: additionally adjusted for alcohol use, the presence of hypertension and diabetes mellitus, serum total cholesterol, serum triglyceride level, HMG-CoA reductase (statin) use.

AHI indicates apnea hypopnea index; OR, odds ratio; SA, sleep apnea; and SpO₂, oxygen saturation.

Carotid-Intima Media Thickness

In minimally adjusted models, greater hypoxemia (≥ 0.64% with SpO₂ <90%) was associated with higher CIMT. However, in fully adjusted models, no significant association between SA or other sleep measures and CIMT were observed (Table 4). In age-group specified analysis adjusted for demographic factors, smoking and BMI, a significant association was observed between a higher percentage of sleep time with SpO₂<90% and CIMT in the younger (β ± SE: 0.032 ± 0.014 mm) but not older participants (0.001 ± 0.013 mm); p-interaction=0.089 (Table 5). This association was slightly attenuated after additionally adjusting for additional cardiovascular risk factors.

Table 4.

Association between Sleep Apnea and Other Sleep Measures and Carotid Intima-Media Thickness

	Model 1 β ± SE	P-value	Model 2 β ± SE	P-value	Model 3 β ± SE	P-value
SA (AHI ≥15 events/h)	0.011 ± 0.010	0.264	−0.002 ± 0.010	0.807	−0.005 ± 0.010	0.642
Arousal index (events/h)	0.0004 ± 0.0004	0.332	0.0001 ± 0.0004	0.793	0.00004 ± 0.0004	0.918
Sleep maintenance efficiency, %	−0.0003 ± 0.0004	0.361	−0.00008 ± 0.0004	0.835	0.000004 ± 0.0004	0.992
≥0.64% Sleep time with SpO2 <90%, %	0.021 ± 0.009	0.018	0.009 ± 0.009	0.346	0.007 ± 0.010	0.444
Total sleep duration, min	−0.00004 ± 0.0001	0.478	0.00000 ± 0.0001	0.997	0.000005 ± 0.0001	0.923
Slow wave sleep, (N3) %	−0.0007 ± 0.0005	0.197	−0.0005 ± 0.0005	0.340	−0.0006 ± 0.0005	0.279

^†Model 1: adjusted for demographic factors (continuous age, race/ethnicity, sex).

Model 2: additionally adjusted for pack-years smoked and body mass index

Model 3: additionally adjusted for alcohol use, the presence of hypertension and diabetes mellitus, serum total cholesterol, serum triglyceride level, HMG-CoA reductase (statin) use.

AHI indicates apnea hypopnea index; SA, sleep apnea; and SpO₂, oxygen saturation.

Table 5.

Association between Sleep Apnea and Other Sleep Measures and Carotid Intima-Media Thickness by Dichotomized Age

	Model 1 β ± SE P-value	P-interaction	Model 2 β ± SE P-value	P-interaction	Model 3 β ± SE P-value	P-interaction
SA (AHI ≥15 events/h)
Age <68	0.034 ± 0.014 0.018	0.041	0.022 ± 0.015 0.125	0.058	0.019 ± 0.015 0.200	0.072
Age ≥68	−0.006 ± 0.014 0.658		−0.014 ± 0.014 0.299		−0.016 ± 0.014 0.242
Arousal index (events/h)
Age <68	0.0009 ± 0.0006 0.115	0.389	0.0006 ± 0.0006 0.335	0.641	0.0005 ± 0.0006 0.341	0.499
Age ≥68	0.0002 ± 0.0005 0.688		0.0002 ± 0.0005 0.724		0.00002 ± 0.0005 0.969
≥0.64% Sleep time with SpO2 <90%, %
Age <68	0.042 ± 0.013 0.001	0.045	0.032 ± 0.01 0.021	0.089	0.028 ± 0.014 0.038	0.106
Age ≥68	0.006 ± 0.013 0.664		0.0007 ± 0.013 0.955		−0.001 ± 0.013 0.928

^†Model 1: adjusted for demographic factors (dichotomized age, race/ethnicity, sex) and interaction term between age and sleep parameter.

Model 2: additionally adjusted for pack-years smoked and body mass index

Model 3: additionally adjusted for alcohol use, the presence of hypertension and diabetes mellitus, serum total cholesterol, serum triglyceride level, HMG-CoA reductase (statin) use.

AHI indicates apnea hypopnea index; SA, sleep apnea; and SpO₂, oxygen saturation.

Analyses evaluating effect modification by sex did not identify significant sex differences although the association between SA and carotid plaque tended to be stronger in men than women (see Supplemental Material). Associations between carotid plaque and SA were stronger in Whites than in other race/ethnic groups. In fully adjusted analyses, significant associations were demonstrated between greater hypoxemia and CIMT in Blacks but not in other groups (p-interaction=0.033). Higher sleep efficiency was associated with CIMT in Hispanics.

DISCUSSION

In this large ethnically/racially diverse community sample, we identified no significant associations between indices of carotid atherosclerosis with measures of disturbed sleep in the overall cohort although several significant associations were observed in individuals and in race-specific analyses. Our results highlight several important findings. First, SA was associated with carotid plaque in individuals younger than 68 years. Higher arousal index also showed a similar pattern of association with carotid plaque in the younger cohort members. Second, decreased oxygen saturation during sleep was the only sleep measure associated with an increase in CIMT, but only in younger or Black individuals. Third, there was no association between self-reported habitual snoring and either measure of carotid atherosclerosis. Fourth, sex did not appear to significantly modify the observed associations. Our findings suggest that: a) Increase in carotid plaque and intima-media thickness may reflect different sleep related stressors. Specifically, carotid plaque may result from the adverse effects of recurrent respiratory disturbances and possible arousal-related sympathetic surges, while CIMT may be increased secondary to stressors associated with hypoxemia, such as inflammation, endothelial dysfunction, hypertension and dyslipidemia;²⁹ b) Associations between sleep disturbances and carotid atherosclerosis tend to be stronger in individuals younger than 68 years compared to older individuals and in Blacks compared to other race/ethnicities although the associations do not vary by sex.

Several epidemiological studies have reported an increased prevalence or incidence of cardiovascular and cerebrovascular disease in individuals with SA.³⁰ The available data, however, raise questions regarding the impact of alternative sleep exposures on risk, as well as suggest heterogeneity in susceptibility to vascular disease across population groups. Although a meta-analysis of published studies reported higher CIMT in patients with obstructive sleep apnea (pooled standardized difference in means 1.40 [lower limit 0.996 to upper limit 1.803]),¹⁸ the authors emphasized limitations of the available data; most studies included less than 100 subjects, have a high degree of bias and heterogeneity, and many did not comprehensively adjust for confounders.

The present study is the largest cohort study to date to examine the association between SA and subclinical carotid atherosclerosis and the first to evaluate the influences of age, sex, and race/ethnicity on this association using rigorously collected sleep measures in a large multiethnic cohort. After adjusting for potential confounders, we observed an association between SA and carotid plaque only among the younger cohort members. We also observed a significant association between CIMT and decreased oxygen saturation during sleep in the younger cohort. These results are consistent with results of the Wisconsin Sleep Study, which found that among individuals with a mean baseline age of 47 years, AHI predicted future carotid plaque presence and score, but not CIMT.³¹ A subsequent study by the same authors found that minimum oxygen saturation, but not other measures of hypoxemia, predicted development of carotid plaques.³² However, minimal oxygen saturation does not fully capture hypoxic burden associated with SA. Our analyses also suggest an association of increasing arousal index and carotid plaque in the younger individuals, which is consistent with a prior study from MESA showing that a higher arousal index was associated with high coronary artery calcium scores.³³ The underlying mechanism may be arousal associated sympathetic activity and blood pressure surges.³⁰ Interestingly, adjustment for hypertension (and other potential intermediate mechanisms) did not substantively alter the odds ratio for this association. However, daytime blood pressure may not reflect the nocturnal blood pressure surges that accompany arousals.

Significant age interaction was previously described between SA and measures of CVD and risk factors. In the Sleep Heart Health Study, SA was associated with systolic/diastolic hypertension in those aged <60 years but not older individuals.²¹ A higher incidence of coronary heart disease in association with SA was also reported in younger compared to older individuals.²⁰ In MESA, left ventricular mass was associated with AHI in the younger cohort members.²² An association between SA and markers of inflammation was also found to be stronger in younger members of MESA.³⁴ The reasons for the observed effect modification by age are not well understood. Possible explanations include age variation in SA characteristics influencing susceptibility to carotid atherosclerosis such as decreased intrathoracic pressure generated with apneic events, and altered chemoreflex and autonomic responses to ventilatory disturbances in older individuals. It is also possible that the weaker associations in older individuals reflect survival bias. Overall, the findings of stronger association between SA and carotid atherosclerosis in younger individuals provide support for the potential benefit of early diagnosis and treatment of SA in preventing cardiovascular disease.

An unexpected finding was the positive association between sleep efficiency and CIMT in Hispanic individuals. Although this may have been spurious, it is possible that this association may reflect the greater propensity of individuals with a high sleep drive to develop cardiovascular disease, as suggested by evidence that sleepiness identifies individuals with SA at risk for cardiovascular disease.³⁵

Although carotid plaque and CIMT are both markers of carotid atherosclerosis, our study points to different associations between indices of disturbed sleep and each measure. These findings are supported by prior data showing CIMT and plaques to reflect different stages and aspects of atherosclerosis, representing different adaptive responses to traditional cardiovascular risk.¹⁵ More recent studies show that carotid plaque presence and carotid plaque score are stronger predictors of CVD and cerebrovascular events than CIMT.^{15, 16, 18, 36}

Strengths of our study include the use of a community-based sample of ethnically diverse women and men. Standardized polysomnography measures and carotid ultrasound reading reduced measurement error, reporting biases, and increased precision. Study limitations include the cross-sectional design, such that one cannot infer causality. The study may be underpowered to detect subgroup interactions. Additional studies powered to evaluate age and race/ethnicity interaction will be of particular interest to further explore the observed association between deceased oxygen saturation during sleep and increased CIMT in Blacks. The use of plaque presence may have less predictive value than quantitative measures of plaques. Finally, despite adjustments for multiple potential confounders, we cannot exclude residual confounding.

In conclusion, our study findings indicate that sleep disturbances are associated with subclinical carotid atherosclerosis in men and women less than 68 years of age. Further, they suggest that the mechanisms underlying the association between SA and carotid atherosclerosis may differ for carotid plaque and CIMT, two biologically and genetically distinct forms of carotid atherosclerotic burden. Finally, race differences suggest the potential for an increased susceptibility of Blacks to the effects of hypoxemia.