Moderate-to-severe obstructive sleep apnea is associated with telomere lengthening

Here, we show that moderate-to-severe obstructive sleep apnea is associated with longer telomeres, independent of age and cardiovascular risk factors, challenging the hypothesis that telomere shortening is a unidirectional process related to age/disease. A better understanding of the mechanisms underlying telomere dynamics may identify targets for therapeutic intervention in cardiovascular aging/other chronic diseases.

Abstract

Obstructive sleep apnea (OSA) is associated with cardiometabolic diseases. Telomere shortening is linked to hypertension, diabetes mellitus, and cardiovascular diseases. Because these conditions are highly prevalent in OSA, we hypothesized that telomere length (TL) would be reduced in OSA patients. We identified 106 OSA and 104 non-OSA subjects who underwent polysomnography evaluation. Quantitative PCR was used to measure telomere length in genomic DNA isolated from peripheral blood samples. The association between OSA and TL was determined using unadjusted and adjusted linear models. There was no difference in TL between the OSA and non-OSA (control) group. However, we observed a J-shaped relationship between TL and OSA severity: the longest TL in moderate-to-severe OSA [4,918 ± 230 (SD) bp] and the shortest TL in mild OSA (4,735 ± 145 bp). Mean TL in moderate-to-severe OSA was significantly longer than in the control group after adjustment for age, sex, body mass index, hypertension, dyslipidemia, and depression (β = 96.0, 95% confidence interval: 15.4–176.6, P = 0.020). In conclusion, moderate-to-severe OSA is associated with telomere lengthening. Our findings support the idea that changes in TL are not unidirectional processes, such that telomere shortening occurs with age and disease but may be prolonged in moderate-to-severe OSA.

NEW & NOTEWORTHY Here, we show that moderate-to-severe obstructive sleep apnea is associated with longer telomeres, independent of age and cardiovascular risk factors, challenging the hypothesis that telomere shortening is a unidirectional process related to age/disease. A better understanding of the mechanisms underlying telomere dynamics may identify targets for therapeutic intervention in cardiovascular aging/other chronic diseases.

obstructive sleep apnea (OSA) is a sleep breathing disorder that affects >20 million United States adults (53, 73). OSA is an independent risk factor for cardiometabolic diseases and is related to worse cardiovascular outcomes. OSA is often diagnosed with hypertension and other cardiovascular diseases such as coronary artery disease, stroke, and atrial fibrillation (13, 29, 43, 66). Cardiovascular diseases are characteristic of aging, and telomeres have been proposed as a marker of disease progression and cardiovascular aging (26, 33, 54, 56).

Telomeres are a repetitive DNA sequence (TTAGGG), which, along with telomere protein complexes, protect the ends of chromosomes from DNA damage, degradation, and fusion (18b, 51). Telomeres shorten with each cell cycle as a result of the incomplete replication of linear DNA (the end-replication problem). Thus, telomeres serve as a checkpoint where the cell cycle is arrested after telomeres reach a critical size. Cells then become senescent or undergo apoptosis (46, 50). Telomeres are therefore postulated to act as a biological clock, reflecting mitotic cell capacity and cumulative exposure to oxidative stress and other environmental stressors over a lifetime (35), and thus to be one of the molecular mechanisms explaining cellular aging.

Numerous studies have investigated the association between telomeres, cardiometabolic risk factors [such as hypertension, diabetes mellitus, physical activity, etc. (31, 58, 70, 75)], and cardiovascular diseases [such as arteriosclerosis (19, 61, 71), ischemic heart disease (24, 32), coronary disease (49, 56), myocardial infarction (10, 18a), and stroke (1, 18a, 21)]. However, the findings are often inconsistent and conflicting showing negative, positive, or no association. Even less clear is the association between telomere length (TL) and OSA, likely because of complex interactions and mechanisms shared between OSA, cardiometabolic risk factors, and cardiovascular disease. Most studies have linked OSA with short telomeres (4, 9, 38, 59, 62, 67), whereas one study has shown telomere elongation (37). Increased oxidative stress and inflammation are among the proposed mechanisms linked to both elevated cardiovascular risk and telomere shortening in OSA (16, 20, 23, 35, 45, 52, 67a, 72). However, the mechanisms underlying telomere shortening have not been completely investigated or confirmed in all studies (4, 62). Importantly, not all studies used polysomnography to identify the presence or absence of OSA (59, 62).

In this cross-sectional exploratory study, we sought to compare TL between non-OSA and OSA groups and examine the association of sleep-related variables with telomere characteristics. We hypothesized that telomere shortening would be accelerated in OSA and thus potentially may be used as a biomarker to assess physiological stress related to OSA and/or disease progression. In this study, we present results that further address the complicated association between OSA and TL.

MATERIALS AND METHODS

Study population.

In this cross-sectional study, we pooled data collected for 210 subjects participating in sleep-related research studies conducted at Mayo Clinic (Rochester, MN) between the years of 2000 and 2007 and for whom extracted DNA was available. All subjects had undergone polysomnography evaluation to confirm or exclude the presence of OSA. Subjects with an apnea-hypopnea index (AHI) of <5 were classified as the control group. Subjects with an AHI of ≥5 were identified as patients with OSA. All OSA subjects were naive to treatment for their sleep disorder. The Mayo Clinic Institutional Review Board approved the study protocol (IRB no. 1872-99), and each participant signed an informed consent form.

Participants with major adverse cardiac events defined as arrhythmia, coronary artery disease, myocardial infarction, and/or heart failure (n = 30) and participants with any type of cancer diagnosed before baseline of sleep study (n = 1) were excluded (Fig. 1) as these conditions have been shown to be associated with TL (33, 63, 69). Demographics and individual and family medical history were collected for all study subjects. Hypertension was defined as self-reported or receiving antihypertensive treatment, dyslipidemia was defined as self-reported dyslipidemia/high lipid concentrations or receiving antihyperlipidemic treatment, smoking was defined as currently smoking, diabetes mellitus was defined as self-reported or as receiving insulin treatment, and depression was defined as receiving medication from antidepressant drug class regardless of a prescription indication. Physical activity was reported as self-reported hours of leisure activity per week.

Fig. 1.

Flowchart of subject inclusion in the study. MACE, major adverse cardiac events; OSA, obstructive sleep apnea; PSG, polysomnographic.

Polysomnography examination.

All participants underwent polysomnographic evaluation at the Mayo Clinic Clinical Research and Trials Unit or the Center for Sleep Medicine. Sleep was scored according to standard criteria (1a, 57). Apneas were defined as complete cessation of airflow lasting ≥10 s, whereas hypopneas were defined as reduction of airflow lasting ≥10 s accompanied by ≥4% decrease in oxygen saturation ($S{p}_{O_2}$). The AHI was computed as a total number of respiratory events per hour of sleep (n/h). Arousal index and mean and minimum $S{p}_{O_2}$ were also determined. On the basis of the AHI, subjects were primarily classified as controls (non-OSA; AHI < 5) or OSA (AHI ≥ 5). In secondary analysis, OSA subjects were grouped as mild OSA (5 ≥ AHI < 15), moderate OSA (15 ≥ AHI < 30), and severe OSA (AHI ≥ 30).

Leukocyte TL measurements.

Peripheral blood samples were collected and stored at −80°C until DNA isolation. DNA was extracted within 1–2 days of blood collection using the PureGene method. Since the DNA extraction method may impact accurate measurements of TL, we ensured that DNA was extracted using the same method in all our study subjects (8, 18). TL was measured by Cawthon’s quantitative PCR (qPCR) method for determining relative TL (14). Briefly, TL qPCR uses two primers designed to hybridize to the telomeric hexamer repeats, and it determines the sample’s telomere-to-single copy gene ratio (T/S) compared with the T/S of a reference DNA sample. The β₂-globin gene on chromosome 11 was used a single copy gene (S) in our study. TL qPCR measurements were performed on an Applied Biosystems Viia 7 machine. For each sample, TL qPCRs were performed in triplicate on a 384-well plate, and failed samples or those with a coefficient of variation of >5% were repeated up to two times. The interassay coefficient of variation was <1% for both telomere and single copy gene primers. Using a reference DNA sample for all TL qPCR runs, a standard curve was generated, which allowed for the reliable determination of TL as T/S (14). T/S was converted into base pairs using the following validated formula: base pairs = (T/S) × 1,470.8 + 7,674.5, as previously described by Boardman et al. (7).

Statistical analysis.

Before any analysis, the distribution of TL was examined, and subjects with TL below 3,842 bp or above 5,873 bp (>3 times the interquartile range above the third quartile or below the first quartile) were excluded as these were regarded as extreme outliers. Pairwise comparisons between OSA groups used a Wilcoxon rank-sum test for continuous variables and Pearson χ²-test for categorical variables. Linear regression evaluated the association between OSA severity and TL and between TL and sleep parameters. In the multivariate analysis, factors known to be associated with cardiovascular risk were included. The first adjusted model controlled for age alone; a second multivariable model controlled for traditional cardiovascular risk factors such as age, sex, body mass index, hypertension, and dyslipidemia; and the third model was additionally adjusted for depression. Information regarding P value and regression estimates (β) to explain the relationship between compared groups is provided. Regression assumptions (homoscedasticity, normality of residuals, and linearity of continuous adjustment factors) were evaluated; no significant violations were detected. A two-tailed P value of <0.05 was considered statistically significant. All analyses were performed with JMP 10.0.0 software (SAS Institute, Cary, NC) and the R program (version 1.0.44).

RESULTS

After exclusion of subjects with major adverse cardiac events (n = 30) and cancer (n = 1) before baseline and extreme outliers (n = 4), the analytic sample included 101 subjects in the non-OSA (control) group and 74 subjects in the OSA group. The baseline characteristics are shown in Table 1.

Table 1.

Characteristics of the study population

	n	Non-OSA	OSA	P Value
Telomere length, bp	175 (101/74)	4,802 (4,711−4,965)	4,824 (4,709−5,009)	0.774
		4,846 ± 190	4,866 ± 224
Age, yr	175 (101/74)	33 (27−43)	47 (38−55)	<0.001
		35 ± 11	48 ± 12
Systolic blood pressure, mmHg	165 (93/72)	122 (112−137)	132 (124−143)	<0.001
		124 ± 16	131 ± 15
Diastolic blood pressure, mmHg	165 (93/72)	74 (66−80)	76 (70−82)	0.057
		73 ± 10	76 ± 9
Heart rate, beats/min	158 (93/65)	69 (62−76)	72 (64−84)	0.031
		70 ± 11	74 ± 14
Body mass index, kg/m2	174 (101/73)	26 (23−28)	32 (28−37)	<0.001
		27 ± 5	32 ± 7
Female sex, n (%)	175 (101/74)	30 (30)	11 (15)	0.022
Diabetes mellitus, n (%)	175 (102/78)	1 (1)	3 (4)	0.180
Dyslipidemia, n (%)	175 (101/74)	5 (5)	12 (16)	0.013
Hypertension, n (%)	175 (101/48)	6 (6)	11 (15)	0.049
Depression, n (%)	175 (101/74)	5 (5)	10 (14)	0.047
Current smoker, n (%)	172 (100/72)	9 (9)	5 (7)	0.627
Physical activity, h/wk	165 (96/69)	3 (0−4)	3 (0−5)	0.922
		2.7 ± 2.1	2.8 ± 2.7
Parental cardiovascular disease history, n (%)	175 (101/74)	25 (25)	32 (43)	0.010
Parental cancer history, n (%)	175 (101/74)	10 (10)	8 (11)	0.845
Arousal index, n/h	175 (101/74)	14 (9−19)	42 (27−63)	<0.001
		15 ± 8	46 ± 22
Apnea-hypopnea index, n/h	175 (101/74)	1 (0−1)	29 (10−51)	<0.001
		1 ± 1	34 ± 27
Minimum blood O2 saturation %	168 (98/70)	91 (88−93)	85 (77−88)	<0.001
		90 ± 5	82 ± 9
Mean blood O2 saturation, %	170 (98/72)	97 (96−98)	94 (93−95)	<0.001
		96 ± 2	94 ± 3

Values are medians (25th−75th percentile) and means ± SD or percentages of number of subjects [n (%)] for each group; n refers to subjects with available data [number of subjects without obstructive sleep apnea (non-OSA)/number of subjects with OSA]. Hypertension was defined as self-reported or receiving antihypertensive treatment, dyslipidemia was defined as self-reported dyslipidemia/high lipid concentrations or receiving antihyperlipidemic treatment, smoking was defined as currently smoking, diabetes mellitus was defined as self-reported or as receiving insulin treatment, and depression was defined as receiving medication from antidepressant drug class regardless of a prescription indication. OSA was defined on the basis of an apnea-hypopnea index ≥5. P values were calculated by Kruskal-Wallis test or χ²-test.

Compared with the non-OSA group, the OSA group was significantly older (P < 0.001) and had higher body mass index (P < 0.001), systolic blood pressure (P = 0.013), and heart rate (P = 0.031) but not diastolic blood pressure (P = 0.057). There were also significantly more men than women, with a higher percentage of women in the non-OSA group than in the OSA group (P = 0.022). The OSA group was also characterized by a higher rate of dyslipidemia (P = 0.013), hypertension (P = 0.049), and depression (P = 0.047); however, there was no difference in smoking status (P = 0.627), diabetes (P = 0.18), or leisure physical activity (P = 0.922). The OSA group reported a higher rate of parental history of cardiovascular disease (P = 0.01), but there was no difference in the rate of parental history of cancer between groups (P = 0.845). As anticipated, the OSA group had a significantly higher arousal index, AHI, and mean and minimum $S{p}_{O_2}$ (P < 0.001).

Comparison of TL between OSA and non-OSA groups showed no statistical difference between groups (P = 0.774). Adjusting for age and additionally for sex, body mass index, hypertension, dyslipidemia, and depression did not change this result (P = 0.536 and P = 0.295, respectively; Fig. 2A). We found no association between parental history of cancer or cardiovascular disease, physical activity, TL, and OSA status. Subjects receiving antidepressants tended to have shorter telomeres than control subjects (regardless of OSA status): 4,789 ± 167 versus 4,861 ± 207 bp (means ± SD, P = 0.160).

Fig. 2.

Telomere length according to obstructive sleep apnea status (A) and severity (B). Boxes represent interquartile range (the third quartile minus the first quartile); whiskers are drawn to the farthest point within 1.5× interquartile range from the box. Horizontal lines indicate medians. P values for comparisons between groups were obtained from models adjusted for age, sex, body mass index, hypertension, dyslipidemia, and depression.

We explored further whether OSA severity, based on the AHI metric, was associated with TL. We divided OSA subjects into three groups: mild OSA (n = 21), moderate OSA (n = 18), and severe OSA (n = 35). We showed that severe OSA and moderate OSA subjects had longer TL than mild OSA subjects (4,906 ± 227 vs. 4,735 ± 145 bp, P = 0.01, and 4,941 ± 241 vs. 4,735 ± 145 bp, P = 0.04, respectively, means ± SD) after adjustment for age. There was no difference in TL between moderate OSA and severe OSA (P = 0.987) and between the control group and any given OSA severity group (Fig. A1). For further analysis, we merged moderate OSA and severe OSA groups into the moderate-to-severe OSA group as there was no difference in TL between these two groups and, clinically, AHI ≥ 15 is associated with an increased cardiovascular risk (66). Table A1 in the appendix shows the subject characteristics across OSA categories.

Linear models used to test the difference between groups showed a J-shaped association between TL and OSA severity: mean TL in the mild OSA group was significantly shorter than in the control group (non-OSA, P = 0.020) and moderate-to-severe OSA group (P < 0.001); the moderate-to-severe OSA group had longer TL than the control group (P = 0.034), even though this group was significantly older than the control group. Adjustment for age (model 1) did not change these main findings; in the model with additional adjustment for recognized cardiovascular risk factors such as sex, body mass index, hypertension, and dyslipidemia (model 2) and in the model additionally adjusted for depression (model 3), the difference in TL between mild OSA and moderate-to-severe OSA and between moderate-to-severe OSA and controls remained statistically significant (Table 2 and Fig. 2B).

Table 2.

Association of telomere length with obstructive sleep apnea categories

	β	−95% Confidence Interval	+95% Confidence Interval	P Value
Unadjusted model (n = 175)				0.002
Mild vs. control	−111.7	−203.6	−19.8	0.020
Moderate to severe vs. mild	183.6	83.2	284.0	<0.001
Moderate to severe vs. control	71.9	5.9	137.9	0.034
Adjusted model 1 (n = 175)				0.005
Mild vs. control	−107.3	−204.0	−10.5	0.031
Moderate to severe vs. mild	184.7	83.9	285.5	<0.001
Moderate to severe vs. control	77.4	4.4	150.4	0.039
Adjusted model 2 (n = 174)				0.046
Mild vs. control	−91.2	−194.1	11.8	0.085
Moderate to severe vs. mild	182.5	78.5	286.2	<0.001
Moderate to severe vs. control	91.4	11.5	171.2	0.026
Adjusted model 3 (n = 174)				0.046
Mild vs. control	−85.9	−189.8	18.0	0.105
Moderate to severe vs. mild	181.9	77.3	286.5	<0.001
Moderate to severe vs. control	96.0	15.4	176.6	0.020

β, β-coefficient. Model 1 was adjusted for age. Model 2 was adjusted for age, sex, body mass index, hypertension, and dyslipidemia. Model 3 was adjusted additionally for depression. P values for comparisons between groups were obtained from the generalized linear model.

We explored the association between TL and sleep parameters in an unadjusted model, a model adjusted for age (model 1), and a model adjusted for age, sex, body mass index, hypertension, and dyslipidemia (model 2; Table A2 in the appendix). We found a positive association between AHI and TL in the fully adjusted model for the entire group (β = 1.5 ± 0.7, means ± SE, P = 0.041). However, we found no association between TL and arousal index, minimum and mean $S{p}_{O_2}$ in the whole group or in the OSA group only (Table A2).

DISCUSSION

The main finding of this study is that OSA is not associated with telomere shortening. Importantly, elongation of telomeres was observed in patients with moderate-to-severe OSA. Additionally, we found a positive association between the AHI and TL for the entire group.

In contrast to our initial hypothesis that TL would be reduced in OSA, we found no difference in TL between the OSA and control (non-OSA) groups. However, exploratory analysis considering OSA severity revealed a J-shaped relationship between TL and OSA: the shortest telomeres in mild OSA and the longest in moderate-to-severe OSA. Importantly, TL of the moderate-to-severe OSA group was even longer than in the control group. This finding is of particular interest as the moderate-to-severe OSA group was significantly older than the control group and would be expected to have shorter TL (47). The difference in TL between moderate-to-severe OSA and controls was independent of age and traditional cardiovascular risk factors such as sex, body mass index, hypertension, dyslipidemia, and depression.

Our results add to the recent studies that have examined TL in health and disease. At the inception of this study, use of polysomnography to evaluate OSA and TL in relatively larger study samples was lacking, and discrepancies in sleep-related factors associated with altered TL existed (4, 9, 38, 67). Since then, several recent well-controlled studies have shown telomere shortening that is more predominant in subjects with moderate and severe OSA (9, 38, 67). Similar to these studies, we showed shorter TL in participants with mild OSA. However, in contrast to these studies, OSA patients with moderate-to-severe OSA in our study showed telomere lengthening. Elongation of TL in moderate-to-severe OSA has been previously demonstrated in the pediatric population (37). The concept of telomere lengthening in patients with moderate-to-severe OSA is not unlikely and needs further investigation. The AHI and oxygen desaturation have been proposed as potential mechanisms underlying telomere shortening in recently published studies (9, 38, 67). In our study, we observed a positive association between AHI and TL but no association with the arousal index or oxygen desaturation. We speculate that a possible effect of OSA (potentially through the downstream cascade of events related to intermittent hypoxemia, blood oxygen desaturation, and/or sleep fragmentation) on TL may be through increased telomerase expression and/or activity. Most human somatic cells present low telomerase expression/activity or are telomerase negative and thus experience telomere shortening with each cell division (15). However, it has been shown that telomerase activity can be modulated by environmental signals such as oxidative stress and inflammation (5, 15, 28). Also, it has been observed that hypoxia upregulated telomerase activity in endothelial cells, and TL seemed to be regulated by a balance between telomere attrition by hypoxia and telomere elongation by enhanced telomerase activity (2, 30). Furthermore, there is evidence that human telomerase can be induced to selectively extend short telomeres by exposure to low pH (27). This finding is of particular interest because acidity is characteristic of both the hypoxic tumor microenvironment and respiratory acidosis in OSA (40). Indeed, telomere lengthening via increased telomerase activity is a hallmark to retain cellular mitotic capacity (15) and several types of cancer (17, 64, 65, 76). Notably, several recent studies have shown an increased cancer risk in OSA patients (11, 12, 39, 41). From this point of view, our results may provide a novel perspective on the biological mechanisms by which OSA may increase the risk of cancer. In the context of the hypothesis that OSA may upregulate telomerase expression/activity, a differential cell response should also be considered. Telomerase activity levels are different in low- and high-turnover cells (15). Therefore, cells with different turnover rates may exhibit differential responses and may manifest opposite effects on TL. Nevertheless, it has been demonstrated that mean TL measured in leukocytes is equivalent to TL in other tissue within the same individual. Future studies are needed to elucidate mechanisms regulating telomere dynamics in OSA.

The nonlinear association of TL with disease progression and/or severity is not unique to our study. The German Chronic Kidney Disease Study demonstrated a U-shaped association between TL and chronic kidney disease duration as opposed to an expected progressive TL shortening with disease duration (43). The authors speculated that chronic oxidative stress and inflammation might activate protective mechanisms leading to telomere lengthening in chronic kidney disease. Upregulated antioxidative mechanisms have been also implicated in OSA (34). It has been also shown that chronic exposure to intermittent hypoxemia may stimulate beneficial physiological effects on the cardiovascular system, an adaptive phenomenon known as intermittent hypoxemia-mediated ischemic preconditioning (44, 48).

Our understanding of telomere dynamics is constantly evolving with growing evidence supporting the concept that changes in TL are not unidirectional processes such that telomere shortening always occurs with age and disease. Several studies have demonstrated both a stable and increased TL over time (3, 36, 68). One longitudinal study showed bidirectional changes in TL in patients with coronary artery disease: TL maintenance was observed in 32% of subjects and TL lengthening in 23% of subjects (25). Importantly, individuals with shorter telomeres maintained or increased their length, whereas those with the longest telomeres experienced a shortening. These studies suggest a threshold effect such that telomere lengthening may be initiated when a critical TL is reached (6).

Some limitations of this study need to be acknowledged. First, this is a cross-sectional study, and we cannot make any inferences or casual statements about mechanisms underlying TL dynamics in OSA; we had no access to blood samples to test for oxidative stress, antioxidant response markers, or telomerase activity. Second, we had no information regarding OSA onset/duration. This may bias our results, as it is conceivable that the observed telomere dynamics may be linked not only to OSA severity but also to exposure duration. Third, although we have adjusted models for traditional confounders relevant to cardiovascular risk, we cannot exclude that other variables related to treatment and lifestyle may affect TL. Specifically, we had no information regarding socioeconomic status or validated assessment of physical activity, anxiety, and depression: we relied on information from past studies, which were not designed primarily to test the association between TL and sleep breathing disorders. Fourth, four subjects were identified as outliers and excluded from analysis. However, it should be noted that these outliers represent 2% of the study population and may represent phenomena or statistical populations that are distinct from those under study. Although the significance of these outliers remains uncertain, it can be assumed that individual TL variability is influenced by unique factors related to genetic background, environment, or other variables that were not included in this study (60). Finally, the relatively modest sample size of each OSA group imposes further limitations on our study. Although we have identified statistically significant associations between OSA severity and TL, wide confidence intervals indicate variability that may reflect individual differences in health, disease, and environmental exposure as described above. Our findings need to be confirmed in larger populations. Indeed, a future confirmation study is needed using data from this study to estimate the power and sample size. Nevertheless, our study has generated new hypotheses and identified areas for further research.

In summary, this exploratory study demonstrated complex telomere dynamics in OSA where mild OSA is related to telomere shortening but moderate-to-severe OSA is associated with telomere lengthening, independent of age. Our findings challenge the hypothesis that telomere shortening is a unidirectional process related to age and age-related disease and highlight several new hypotheses. A better understanding of the mechanisms underlying telomere dynamics may identify potential targets for therapeutic intervention in cardiovascular aging and other chronic diseases.

GRANTS

This work was supported by National Institutes of Health Grants HL-65176, HL-114676, and HL-114024 (to V. K. Somers and P. Singh) and CA-2024013 (to L. A. Boardman), American Heart Association Grants 16POST30260005 (to K. Polonis), 16POST27210011 (to C. Becari), and 16SDG27250156 (to N. Covassin), and National Council for Scientific and Technological Development Brazil Grant 203802/2014-4 (to C. Becari).

DISCLOSURES

V. K. Somers: grant support from Philips Respironics Foundation (gift to Mayo Foundation); consultant for Respicardia, ResMed, Sorin Incorporated, U-Health, GlaxoSmithKline, Rhonda Grey, Dane Garvin, Philips Respironics, Biosense Webster; working with Mayo Health Solutions and their industry partners on intellectual property related to sleep and cardiovascular disease.

AUTHOR CONTRIBUTIONS

K.P., V.K.S., B.R.D., L.A.B., and P.S. conceived and designed research; B.R.D. and R.A.J. performed experiments; K.P. and P.J.S. analyzed data; K.P., V.K.S., C.B., N.C., and P.S. interpreted results of experiments; K.P. prepared figures; K.P. drafted manuscript; K.P., V.K.S., C.B., N.C., P.J.S., B.R.D., K.N., L.A.B., and P.S. edited and revised manuscript; K.P., V.K.S., C.B., N.C., P.J.S., B.R.D., R.A.J., K.N., L.A.B., and P.S. approved final version of manuscript.

APPENDIX

Box plots of TL across OSA severity groups are shown in Fig. A1. Characteristics of the study population across categories of OSA severity are shown in Table A1, and the association between TL and sleep parameters is shown in Table A2.

Fig. A1.

Box plots of telomere length across obstructive sleep apnea severity groups. P values for comparisons between groups were obtained from the age-adjusted model. Boxes represent interquartile range (the third quartile minus the first quartile); whiskers are drawn to the farthest point within 1.5× interquartile range from the box. Horizontal lines indicate medians.

Table A1.

Characteristics of the study population across categories of OSA severity

	n	Control (non-OSA)	Mild OSA	Moderate-to-Severe OSA	P Value
Telomere length, bp	175 (101/21/53)	4,802 (4,711−4,965)a	4,719 (4,641−4,812)b	4,902 (4,733−5,081)a	0.002
		4,846 ± 190	4,735 ± 145	4,918 ± 230
Age, yr	175 (101/21/53)	33 (27−43)a	45 (37−51)b	48 (39−57)b	<0.001
		35 ± 11	45 ± 10	47 ± 13
Systolic blood pressure, mmHg	165 (93/20/52)	122 (112−137)a	124 (114−139)b	134 (129−143)b	<0.001
		124 ± 16	124 ± 18	134 ± 13
Diastolic blood pressure, mmHg	165 (93/20/52)	74 (66−80)	75 (69−84)	76 (71−82)	0.152
		73 ± 10	76 ± 11	76 ± 8
Heart rate, beats/min	158 (93/18/47)	69 (62−76)	74 (67−88)	72 (64−83)	0.079
		70 ± 11	78 ± 18	74 ± 12
Body mass index, kg/m2	174 (101/20/53)	26 (23−28)a	31 (25−39)b	32 (29−36)b	<0.001
		27 ± 5	32 ± 8	33 ± 6
Female sex, n (%)	175 (101/21/53)	30 (30)	5 (24)	6 (11)	0.004
Diabetes mellitus, n (%)	175 (101/21/53)	1 (1)	2 (10)	1 (2)	0.06
Dyslipidemia, n (%)	175 (101/21/53)	5 (5)	4 (19)	8 (15)	<0.001
Hypertension, n (%)	175 (101/21/53)	7 (7)	3 (14)	11 (21)	0.040
Depression, n (%)	175 (101/21/53)	5 (5)	3 (14)	7 (13)	0.137
Current smoker, n (%)	175 (101/21/53)	9 (9)	1 (5)	4 (8)	0.846
Physical activity, h/wk	165 (96/17/52)	3 (0−4)	4 (0−5)	2.5 (0−5)	0.928
		2.7 ± 2.1	2.5 ± 2.4	2.8 ± 2.8
Parental cardiovascular disease history, n (%)	175 (101/21/53)	25 (25)	6 (29)	26 (49)	0.010
Parental cancer history, n (%)	175 (101/21/53)	10 (10)	2 (10)	6 (11)	0.956
Arousal index, n/h	175 (101/21/53)	14 (9−19)a	27 (20−41)b	47 (30−69)c	<0.001
		15 ± 8	33 ± 18	51 ± 22
Apnea-hypopnea index, n/h	175 (101/21/53)	1 (0−1)a	6 (5−9)b	40 (23−58)c	<0.001
		1 ± 1	7 ± 2	44 ± 26
Minimum blood O2 saturation, %	168 (98/20/50)	91 (88−93)a	89 (84−90)b	83 (75−86)c	<0.001
		90 ± 5	88 ± 4	80 ± 10
Mean blood O2 saturation, %	170 (98/20/52)	97 (96−98)a	95 (92−95)b	94 (92−95)c	<0.001
		96 ± 2	95 ± 3	93 ± 3

Values are medians (25th–75th percentile) and means ± SD or percentages of numbers of subjects [n (%)] for each group; n refers to those subjects with available data [number of control subjects/number of subjects with mild obstructive sleep apnea (OSA)/number of subjects with moderate-to-severe OSA]. Hypertension was defined as self-reported or receiving antihypertensive treatment, dyslipidemia was defined as self-reported dyslipidemia/high lipid concentrations or receiving antihyperlipidemic treatment, smoking was defined as currently smoking, diabetes mellitus was defined as self-reported or as receiving insulin treatment, and depression was defined as receiving medication from antidepressant drug class regardless of a prescription indication. P values were calculated by Kruskal-Wallis test or χ²-test.

^a,b,cLevels not connected by same letter are significantly different at P < 0.05 (nonparametric comparisons for each pair using the Wilcoxon method).

Table A2.

Association between telomere length and sleep parameters

	Unadjusted		Model 1		Model 2
	All	OSA	All	OSA	All	OSA
Apnea-hypopnea index
n/h	1.1 ± 0.6	1.6 ± 0.9	1.3 ± 0.7	1.6 ± 1.0	1.5 ± 0.7	1.5 ± 1.0
P value	0.077	0.091	0.069	0.097	0.041	0.151
Arousal index
n/h	0.9 ± 0.7	1.2 ± 1.2	1.1 ± 0.8	1.2 ± 1.2	1.5 ± 0.9	0.8 ± 1.4
P value	0.195	0.328	0.167	0.346	0.096	0.538
Minimum blood O2 saturation
%	−0.3 ± 2.0	−0.8 ± 3.0	0.03 ± 2.1	−0.7 ± 2.3	−0.5 ± 2.4	−0.3 ± 3.5
P value	0.875	0.804	0.988	0.823	0.841	0.934
Mean blood O2 saturation
%	1.5 ± 6.1	8.0 ± 9.4	2.3 ± 6.4	8.0 ± 9.5	1.5 ± 6.8	4.9 ± 9.9
P value	0.803	0.400	0.717	0.401	0.823	0.627

Values are β-coefficients ± SE with the corresponding P value. OSA, obstructive sleep apnea. Model 1 was adjusted for age. Model 2 was adjusted for age, sex, body mass index, hypertension, and dyslipidemia.