Author + information
- Received March 30, 2016
- Revision received August 26, 2016
- Accepted September 1, 2016
- Published online May 15, 2017.
- Dileep Raman, MDa,
- Farhad Kaffashi, PhDb,
- Li-Yung Lui, MA, MSc,
- William H. Sauer, MDd,
- Susan Redline, MD, MPHe,f,
- Peter H. Stone, MDe,
- Peggy M. Cawthon, PhD, MPHc,g,
- Katie L. Stone, PhDc,g,
- Kristine E. Ensrud, MD, MPHh,i,
- Sonia Ancoli-Israel, PhDj,
- Kenneth A. Loparo, PhDc,
- Reena Mehra, MD, MSk,∗ (, )
- MrOS Study Group
- aSleep Disorders Center, Neurological Institute, Cleveland Clinic, Cleveland, Ohio
- bDepartment of Electrical Engineering and Computer Science, Case Western Reserve University, Cleveland, Ohio
- cResearch Institute, California Pacific Medical Center, San Francisco, California
- dUniversity of Colorado School of Medicine, Aurora, Colorado
- eDepartment of Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts
- fDepartment of Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, Massachusetts
- gDepartment of Epidemiology and Biostatistics, University of California, San Francisco, California
- hDepartment of Medicine, Division of Epidemiology and Community Health University of Minnesota, Minneapolis, Minnesota
- iDepartment of Medicine and Division of Epidemiology and Community Health, Minneapolis VA Health Care System, Minneapolis, Minnesota
- jDepartment of Psychiatry, University of California-San Diego, La Jolla, California
- kCleveland Clinic Lerner College of Medicine, Cleveland, Ohio
- ↵∗Address for correspondence:
Dr. Reena Mehra, Cleveland Clinic Foundation, Sleep Center, Neurologic Institute, 9500 Euclid Avenue, Cleveland, Ohio 44195.
Objectives The authors hypothesized that polysomnogram-based heart rate variability autonomic function biomarkers are associated with incident atrial fibrillation (AF) and these associations are modified by measures of sleep-disordered breathing.
Background Autonomic dysfunction contributes to AF.
Methods A total of 2,350 participants of a multicenter prospective study (Outcomes of Sleep Disorders in Older Men Study) without baseline AF underwent sleep studies with incident adjudicated AF follow-up (8.0 ± 2.6 years). Cox proportional hazard models were used to analyze sleep study–electrocardiogram spectral heart rate variability indices (low- and high-frequency power [LF/HF]) and time domain indices (mean of normal to normal beats and short- and long-term variability) and premature atrial contractions and incident AF (hazard ratio and 95% confidence interval). Statistical interactions between heart rate variability and sleep-disordered breathing were examined. Models were adjusted for age, race, body mass index, waist circumference, cardiac medications, comorbid diseases, alcohol use, and study site.
Results Lower LF/HF and lower LF were associated with higher AF incidence (LF/HF Q1 vs. Q4: 1.46; 1.02 to 2.08; LF Q1 vs. Q4: 1.46; 1.02 to 2.10). Higher short- and long-term variability was associated with an increased risk of AF (p trend = 0.028). The highest premature atrial contractions quartile had a 3-fold increased AF risk (2.99; 1.94 to 4.62) compared with the lowest quartile. A significant interaction of obstructive apnea was observed in the LF-AF relationship (p = 0.045).
Conclusions Sleep-related reduced sympathovagal balance (LF/HF) and increased atrial ectopy are independently associated with future AF, a relationship modified by obstructive apnea.
Sleep-disordered breathing (SDB), a highly prevalent disorder, is characterized by repetitive upper airway collapse and attendant intermittent hypoxia, intrathoracic pressure alterations, and autonomic nervous system fluctuations. SDB prevalence is increasing and is 2- to 4-fold more common in men (1) and upward of 20% in elderly men (2). SDB-related autonomic dysfunction likely contributes to atrial fibrillation (AF) development (3), the latter associated with considerable morbidity and mortality. Similar to SDB, AF prevalence is expected to rise given the increased percentage of the aged population (4).
Enhanced vagal activity during apnea and hypopnea events is punctuated by sympathetic nervous system activation, thereby creating conditions for electrical remodeling and cardiac arrhythmogenesis. Muscle sympathetic nervous system activation is augmented in sleep apnea, findings that persist during wakefulness and improve with treatment (5). In a canine model, right ganglionated plexus ablation inhibited apnea-induced AF (3). Thus, SDB may represent a novel target for AF prevention and treatment strategies as underscored in a recent report (6).
Furthermore, there are 1.1 million polysomnograms (PSGs) performed annually in the United States alone, mainly to test for SDB (7). Given the high SDB prevalence and underdiagnosis, efforts are underway to increase recognition and enhance efficiency via the use of home sleep apnea testing resulting in further increases in sleep testing volumes. Sleep testing, a potential untapped resource, may provide physiological signatures to predict adverse health outcomes, such as AF. Although there is a high throughput of sleep studies that include standard respiratory channels in PSG and home sleep apnea testing, there are no consistent recommendations to include the electrocardiogram (ECG) signal (8).
We investigate whether PSG-based cardiac electrophysiological indices, heart rate variability (HRV), and atrial ectopy, the latter recognized to predict AF (9), are associated with increased risk of AF development. We focus on ECG measures most likely to represent biological correlates/precursors of AF development i.e., HRV as a marker of autonomic function and premature atrial contractions [PAC] burden as an indication of pulmonary vein trigger potential. We leverage a large community-based cohort with careful collection of objective sleep measures in a group of older men at increased risk for AF and its attendant morbidity. Given autonomic fluctuations that accompany apneas and hypopneas, we also examined the effect modification of SDB indices.
Participants and study design
The MrOS (Outcomes of Sleep Disorders in Older Men Study) is a prospective, observational ancillary study of the Osteoporotic Fractures in Men Study. In the parent MrOS study, 5,994 community-dwelling men aged 65 and older able to ambulate without assistance, and without history of bilateral hip replacement, were initially enrolled in 2000 to 2002 at 6 centers (Birmingham, Alabama; Minneapolis, Minnesota; Monongahela Valley near Pittsburgh, Pennsylvania; Palo Alto, California; Portland, Oregon; and San Diego, California). The MrOS study design, methods, and demographics were previously published (10–12).
Of the 3,135 MrOS Sleep Study participants who completed a repeat visit from December 2003 through March 2005, a total of 179 did not participate in polysomnography because of refusal or treatment of SDB and 45 men had a failed sleep study (1.5%). Of the 2,911 participants with a valid sleep study, 121 had pacemakers or poor ECG quality, 136 had prevalent AF identified on the baseline PSG, 266 did not have 5 continuous min of ECG data without artifact or ectopy, and 38 men were without data on adjudicated AF events. Thus, the final analytic sample included 2,350 participants with a mean duration of follow-up 8.0 ± 2.6 years (Figure 1). Each site and the study coordinating center received ethics approval from their institutional review board. Written informed consent was obtained from all participants.
Electrophysiological data analysis
HRV signal processing analyses were performed from the lead II ECG signal (sampled at 250 Hz) of the PSG. Pre-processing of the ECG channel was conducted via sequential automated and subsequent manual scoring involving visual inspection of the PSG ECG data using the Somte software (Compumedics, Inc., Melbourne, Australia). Automated analyses were applied to the ECG data with subsequent manual review (high interobserver and intraobserver reliability [intraclass correlation coefficient (ICC) = 0.98 to 0.99] for ventricular and atrial ectopic beats ) to identify normal sinus beats, ventricular ectopic beats (premature beat with QRS duration >120 ms), atrial ectopic beats (premature beat with QRS duration <120 ms), and artifact. PACs were identified per h of sleep.
To compute HRV analyses, as per standard recommendations (14), all 5-min segments (standard duration to ensure stationarity of the ECG time-series) from the PSG ECG recordings of continuous nonoverlapping ECG (i.e., without ectopy or artifact as informed by pre-processed ECG data) were identified; only normal sinus beats were used for analyses (15). Artifact rejection algorithms intrinsic to the HRV analysis were also used to exclude those segments with heart rate lower than 30 beats/min or >180 beats/min or instantaneous heart rate change exceeding 80 beats/min between 2 consecutive R-R intervals.
Time series measures analyzing normal-normal R-R intervals included conventional time domain HRV measures, such as mean of the normal-normal intervals and those derived from the Poincare′ plot. Here normal-normal refers to intervals between normal heartbeats resulting from sinus node depolarizations that are detected between adjacent QRS complexes of continuous artifact-free ECG. Specific parameters from the Poincare′ analysis include short- and long-term variability measures (STV, LTV). In the analysis, consecutive beat-to-beat data are analyzed as a scattergram and the distribution of points along orthogonal directions that define the minor and major axes of a hypothetical ellipse that fits the data represent the STV and LTV.
Frequency-based or spectral HRV measures were also considered as primary predictors and analyzed using approaches for the analysis of nonuniformly based R-R interval data for spectral analyses of the R-R time series (16). Using this approach, the normalized low-frequency power (LF) (from 0.04 to 0.15 Hz) and the normalized high-frequency power (HF) (from 0.15 to 0.4 Hz) were calculated. The activity in the HF band is considered to be primarily caused by parasympathetic activity of the sinoatrial node, and the LF region is generally considered to reflect sympathetically mediated activity (17). The LF to HF ratio (LF/HF) is often used as a measure of sympathovagal balance.
Adjudicated AF and follow-up
Participants were queried every 4 months about cardiovascular events requiring hospitalization or emergency department visit by mailed questionnaire and/or telephone contact (>99% response rate). AF events were then centrally adjudicated by a board-certified cardiologist using a pre-specified standard protocol using medical records and supporting documentation. AF events were defined using similar criteria from prior studies (18–20) as those resulting in symptoms or an emergency department visit, hospitalization and/or prolongation of a hospitalization, or a procedure directly attributable to AF.
Symptoms considered for adjudication included fatigue, palpitations, lightheadedness, pre-syncope, syncope, chest pain, or dyspnea. Documentation required for an adjudicated AF event included 1 or more of the following: emergency medical services notes and/or rhythm strips, ECG (including stress testing), in-hospital telemetry, ambulatory ECG (Holter monitor and/or event monitor), pacemaker or defibrillator telemetry (for those patients with a device already implanted), or invasive cardiac electrophysiology testing. AF and atrial flutter events included either of these tachycardias and any cardioversion procedures to restore normal sinus rhythm. As part of the triannual postcard contact, participants were asked about SDB treatment. Among those without AF, 575 died and 63 terminated during the follow-up period and were right censored.
An unattended 14-channel home PSG (Safiro, Compumedics, Inc.) was performed within 6.9 ± 15.8 days from the MrOS Sleep visit and was set up by certified staff in the participant’s home for 1 night. The sleep study involved use of C3/A2 and C4/A1 electroencephalograms, bilateral electrooculograms, a bipolar submental electromyogram, thoracic and abdominal respiratory inductance plethysmography, airflow (by nasal-oral thermocouple and nasal pressure cannula), finger pulse oximetry, lead I ECG (250 Hz), body position, and bilateral leg movements. Apnea was defined as complete or near complete cessation of airflow for more than 10 s. The event was categorized as obstructive if effort persisted on thoracoabdominal inductance channels or as central if there was no effort detected. Hypopneas were scored if clear reductions in breathing amplitude (at least 30% below baseline breathing) occurred and lasted >10 s with a drop in arterial saturation of 3% or more (21). The interscorer reliability for the apnea hypopnea index (AHI) was high (ICC = 0.99) (22). SDB severity was defined by the total number of apneas and hypopneas. The obstructive AHI (OAHI) was calculated as the number of obstructive apneas and hypopneas associated with a ≥3% desaturation per h of sleep, and central apnea index (CAI) was calculated as the number of central apneas per h of sleep.
Questionnaires were completed by all participants at the sleep visit. Demographic information, personal habit, and medical history data were collected. Participants were asked to provide all current medications used within the last 30 days (23); prescription and nonprescription medication information was collected, entered into an electronic database, and matched to its ingredients based on the Iowa Drug Information Service Drug Vocabulary (College of Pharmacy, University of Iowa, Iowa City, Iowa) (23). Cardiovascular medications included calcium- channel blockers, nonophthalmic beta-blockers, cardiac glycosides, or antiarrhythmic medications (cardiac sodium channel blockers and potassium channel blockers) (23). Body mass index (kg/m2) was calculated from body weight measured with standard balance beam or digital scale calibrated with standard weights and height measured with a wall-mounted Harpenden stadiometer. Waist circumference was measured (cm). The presence of pacing during the recording was ascertained by examination of the PSG ECG recording. Cholesterol was measured an average of 3.4 years earlier during the MrOS baseline visit using a Roche COBAS Integra 800 automated analyzer that was calibrated daily (Roche Diagnostics Corp., Indianapolis, Indiana). Total cholesterol (mg/dl) was calculated as follows: high-density lipoprotein (mg/dl) + low-density lipoprotein (mg/dl) + 0.2 · triglycerides (mg/dl). Cardiovascular disease was defined by history of myocardial infarction, coronary angioplasty, stroke, and/or coronary artery bypass graft surgery by participant self-report of physician diagnoses.
Participant characteristics were summarized as mean ± SD or n (%) and were compared using chi-square tests for categorical variables and Student t test for continuous variables. HRV indices were expressed by quartile with frequency-based HRV indices (LF, HF, LF/HF) considered as primary predictors given generally accepted use as indices of sympathovagal balance. Secondary HRV measures considered include time domain mean of the normal-normal intervals, STV, and STV/LTV indices. Cox proportional hazard models were used to determine the risk of incident AF across HRV quartiles. A test of trend was performed across the quartiles to assess for monotonic relationships of HRV and atrial ectopy measurements and incident AF. PACs were also examined given data implicating atrial ectopy as a harbinger of AF development (9). Unadjusted models were performed and multivariable models adjusted for age, race, body mass index (kilogram per square meter), waist circumference (centimeter), self-reported medical history (including the following considered as individual covariates: cardiovascular disease, heart failure, hypertension, diabetes mellitus, chronic obstructive pulmonary disease), antiarrhythmic/beta-blocker/calcium-channel blocker medications, total cholesterol, alcohol use, and study site. Results are presented as hazard ratios (HRs) with 95% confidence intervals (CIs). P values for trend are presented reflecting the trend of risk of incident AF across quartiles of HRV measure. Secondary analyses were conducted testing for the statistical interaction of SDB indices (AHI ≥15, OAHI ≥15, CAI ≥5) with HRV variables and stratification was performed. Interaction terms were considered significant if p < 0.10. To assess robustness of analyses, we performed separate sensitivity analyses excluding: 1) participants with second- or third-degree atrioventricular block (to address confounding by conduction delay arrhythmia); 2) participants with <5 usable HRV epochs of data; and 3) participants using positive airway pressure therapy. All significance levels reported are 2-sided, and all analyses were conducted using SAS version 9.4 (SAS Institute Inc., Cary, North Carolina).
Of the 2,350 participants, over the 8.0 ± 2.6 year follow-up period, incident adjudicated AF was observed in 269 (11.4%) subjects (Table 1). Of all participants, 89.7% were white, the average age was 75.8 ± 5.3 years, and overall, participants were nonobese (27.1 ± 3.7 kg/m2). A significant proportion had cardiovascular disease (26.1%) and was taking cardiovascular medications (37.2%). These percentages were significantly higher in the incident AF group (40.3% and 57.3% for those without and with AF, respectively). Those with AF were also more likely to have hypertension. Overall, a median of 56 HRV epochs of 5 min (range, 1 to 135 min) were identified for analyses.
Electrophsyiological indices and incidence of AF
Proportional hazards assumptions were evaluated statistically and determined to be satisfied for all models. In multivariable adjusted analyses, those in the lowest LF quartile were 46% more likely to develop AF compared with the reference group (quartile 4). Similarly, participants in the lowest LF/HF quartile had 46% higher risk of incident AF compared with the highest quartile (Table 2). In the unadjusted analyses for the frequency domain HRV indices, lower LF and LF/HF were associated with increasing risk of incident AF (all p trend ≤0.001), and the association remained statistically significant for HF and LF/HF in multivariable adjusted analyses (p trend = 0.043 and 0.021, respectively). For the time domain HRV measures, there was a significant relationship of increasing mean of the normal-normal intervals, STV and STV/LTV, and higher risk of AF (p trend = 0.011, 0.003, and 0.0001, respectively); however, in multivariable adjusted analyses significance persisted for STV/LTV only (p trend = 0.028) (Table 3). A statistically significant relationship with a strong magnitude of association of PACs and incident AF was observed with an approximate 3-fold increased risk of AF in the highest quartile compared with the reference quartile 1 (HR: 2.99; 95% CI: 1.94 to 4.62). Moreover, compared with quartile 1, quartiles 2 and 3 of PACs were associated with a 2.2-fold (HR: 2.19; 95% CI: 1.31 to 3.43) and 2.5-fold (HR: 2.50; 95% CI: 1.62 to 3.87) increased risk of incident AF compared with the reference quartile, respectively (Figure 2).
Tests of interaction were used to analyze how SDB modified the relationship of HRV indices and incident AF. There was a trend toward a significant interaction between LF and SDB (defined as AHI ≥15) and a significant interaction of obstructive sleep apnea (defined as OAHI ≥15) and LF in relation to incident AF (p = 0.060 and p = 0.045, respectively). In multivariable adjusted stratified analyses, a 35% increased risk of AF per 1-SD decrease in LF was observed in those with AHI ≥15 (HR: 1.35; 95% CI: 1.09 to 1.68) and OAHI ≥15 (HR: 1.34; 95% CI: 1.08 to 1.66), a finding not observed in those with AHI <15 (HR: 0.99; 95% CI: 0.83 to 1.18) or OAHI <15 (HR: 1.00; 95% CI: 0.84 to 1.20) (Figure 3). Similarly, a trend toward a statistically significant interaction was observed between HF and central sleep apnea (defined as CAI ≥5); p = 0.06 with multivariable adjusted stratified analyses showing more pronounced relationships of increases in HF and increased AF in those with CAI ≥5 (2.12 per SD increment in HF, 1.17 to 3.85) compared with CAI <5 (1.11 per SD increment in HF, 0.97 to 1.28). There was no significant statistical interaction of PAC frequency and SDB indices relative to incident AF.
After excluding 51 (2.2%) men with second- or third-degree atrioventricular block, there were no substantive changes in the results. Separate analyses, excluding those participants using continuous positive airway pressure (n = 20; 0.9%) also did not appreciably alter results. After excluding 160 (6.8%) studies with <5 epochs of usable HRV data, findings were strengthened in terms of magnitude of association and the trend of increasing AF across HF quartile remained statistically significant in multivariable adjusted analyses with 54% increased AF in the highest quartile (HR: 1.54; 95% CI: 1.06 to 2.24) compared with the lowest quartile (p trend = 0.010).
In this multicenter, community-based cohort of older men, we observe that PSG-based derived frequency domains of HRV predict increasing AF incidence. A progressive reduction in LF/HF (i.e., a reflection of lower sympathetic to parasympathetic activity) was associated with increased adjudicated AF incidence over 8-year follow-up after consideration of confounders, such as obesity, cardiac disease, and cardiac medications. Men in the lowest LF quartile (i.e., indicative of low level of sympathetic activation) had a 46% increased adjusted risk of AF compared with the highest LF quartile. Monotonic increases in time domain HRV indices, STV and STV/LTV, demonstrated significant associations with increased AF risk across increasing quartile. Additionally, increasing quartile of PACs per h of sleep was associated with increased adjusted AF risk with a near 3-fold increased risk of AF in the highest compared with the lowest quartile. SDB indices of obstructive sleep apnea modified the relationship of LF and incident AF (i.e., 33% increased AF risk relative to reduction in LF in severe obstructive apnea; findings not observed in lesser apnea). This work addresses research priorities outlined in the Heart Rhythm Society Atrial Prevention guideline to examine SDB and PSG determinants of AF development (6).
Autonomic activity is increased in patients with SDB. It has been previously shown that alterations in autonomic regulation modulates SDB-related atrial arrhythmogenesis (5). HRV is a widely used tool to measure autonomic activity. Among the frequency-dependent spectral variables, HF (0.15 to 0.40 Hz) is thought to reflect parasympathetic tone and the LF (0.04 to 0.15 Hz) provides combined information of the parasympathetic and the sympathetic nervous system (14). However, we recognize that this may represent an oversimplification and may not accurately reflect all aspects of the 2 limbs of the autonomic nervous system, but rather tend to reflect the overall sympathovagal balance. PSG-derived HRV presents an easily obtainable and analyzable source of continuous autonomic activity monitoring. Because the raw data are already recorded, only further software-based analysis is required with use of analytical techniques, which are already well-established.
Previous work using 24-h Holter monitoring in the Cardiovascular Health Study shows that PAC count confers an additional risk for AF beyond traditional risk factors: a median hourly PAC count at baseline was significantly higher in participants with incident AF (5.3 beats/h) compared with without incident (1.8 beats/h) (9). These results are consistent with our findings of a relatively low threshold of PAC burden conferring increased AF risk (9). The novelty of the current findings lies in identifying AF predictive value of PAC frequency from nocturnal ECG recordings as opposed to 24-h continuous ECG monitoring. Our results suggest the possible use of nocturnal PSG-derived electrophysiological markers to verify PAC frequency as an AF risk without the requirement of longer duration ECG monitoring. Because PACs in pulmonary veins can result in AF triggering and ablation of PACs may reduce AF recurrence (24), the ability of PSG-based ECG monitoring may play a role in identification of those most susceptible to target for AF prevention and treatment.
As the evidence for the link between SDB and AF (and other arrhythmias) continues to grow, we will require more robust and readily available tools to quantify and improve prediction of AF in SDB. PSG-derived HRV variables may provide an opportunity to allow for risk quantification and stratification, although not systematically examined in the current work. Use of these markers may provide an early warning for AF risk and possibly the ability to tailor effective treatment to mitigate risk in susceptible individuals. Normalization of these indices reflecting improvement of autonomic dysfunction may serve as a physiological marker of SDB treatment compliance and efficacy particularly as we currently rely on positive airway pressure machine derived proof of compliance with SDB therapy and symptom-based proof of efficacy. This, however, remains speculative and requires prospective confirmation. Separation of HRV indices based on sleep stage may also provide useful data as it relates to SDB. Alternatively, normalization of HRV and PAC metrics may predict long-term effectiveness of AF therapies.
Strengths of this study include its large sample size generalizable to healthy, community-dwelling older men (prone to develop AF); the prospective and longitudinal nature; the rigorous HRV analysis using standard methodology; use of SDB cutoffs relevant to current practice; careful consideration of confounding factors; and multiple sensitivity analyses to exclude influence of overt conduction delay.
The limitations are that this study is not generalizable to women or young men, infirmed older men, or nonwhites. The observational design precludes definitively excluding the possibility of residual confounding. Despite biological plausibility, the observational design precludes ascertainment of definitive causal conclusions of sleep apnea–related HRV resulting in increased AF risk. Furthermore, other aspects of SDB-related pathophysiology, such as altered cardiac substrate (e.g., atrial remodeling and scarring) (25) resulting in increased AF, were not examined. Finally, although HRV analyses are a widely used tool in cardiology because of sound reproducibility, noninvasive nature, and data supporting these indices as a cardiac disease prognostic marker (26), limitations should be recognized, including assumptions of heart rate, respiration, and complexities of nonlinear and sometimes reciprocal relationships of sympathetic and parasympathetic nerve activity (27).
In summary, a progressive reduction of a surrogate of sleep-related sympathovagal balance (LF/HF) derived from PSG-based ECG data represents an independent risk factor for AF development given preservation of point estimate strength after accounting for a multitude of confounders. These results suggest that enhanced vagal tone detected by PSG-based electrophysiological indices represents a potential forecasting AF biomarker. Furthermore, obstructive sleep apnea, a disorder accompanied by enhanced parasympathetic activity during the apneic and hypopneic events, seems to modulate the relationship of LF/HF and incident AF. This is consistent with the biological basis of obstructive sleep apnea–related autonomic alterations resulting in AF possibly via electrical remodeling. Frequency of PACs during 24-h continuous ECG monitoring has been linked to increased AF and mortality. Similarly, we observe association of PAC frequency identified by limited PSG-based ECG monitoring during sleep and incident AF. These findings have potential implications in terms of using PSG-based physiologic signatures for risk stratification and AF preventative or therapeutic targets thereby providing basis for further study. Future investigation should assess whether subtle cardiac conduction abnormalities may distort the expected association of electrophysiological biomarkers of sympathetic/parasympathetic balance and AF development. The contribution of sleep state influences with inherent autonomic characteristics also represents an area of future investigation. The impact of state- or stage-specific alterations in PSG ECG-based biomarker autonomic physiology has been described (28); however, these interrelationships as they pertain to AF risk should also be investigated. Furthermore, it remains to be seen whether reversal of SDB pathophysiology with standard therapy (i.e., positive airway pressure), or otherwise, attenuates SDB-modulated alterations in HRV parameters and alters outcome as it relates to incident AF risk.
COMPETENCY IN MEDICAL KNOWLEDGE: In the current era of enhanced recognition of the importance of sleep-disordered breathing, the conduct of sleep studies, particularly home sleep apnea testing (the latter, unlike in-laboratory sleep studies, not standardly conducted with ECG monitoring), are in parallel increasing and result in more than 1 million sleep studies performed annually. Continuous ECG monitoring conducted as part of sleep study monitoring provides a unique opportunity to leverage embedded physiological signatures that predict adverse outcomes, such as atrial fibrillation. The current work identifies heart rate variability measures of autonomic function and atrial ectopy as a surrogate of pulmonary vein triggers as predictors of incident atrial fibrillation in a cohort of older men, the latter a group vulnerable to development of actionable atrial fibrillation. Findings suggest the potential utility of standards including the ECG signal as part of sleep apnea testing as a salient resource for atrial fibrillation risk stratification.
TRANSLATIONAL OUTLOOK: Further work is needed to identify specific heart rate variability or atrial ectopy thresholds that confer the greatest increased risk of atrial fibrillation development. Although the current work provides the platform to pursue this investigation, findings require validation and extrapolation in independent data. Refinement of the influence of apnea-related breathing disturbances and delineation of cardiac electrophysiological data across sleep stage is also of interest. Future investigation should be focused on the impact of reversal of sleep-disordered breathing–related autonomic physiology on alteration of heart rate variability and atrial ectopic parameters as it relates to atrial fibrillation risk.
The MrOS Study is supported by National Institutes of Health (NIH). The following institutes provide support: the National Institute of Arthritis and Musculoskeletal and Skin Diseases; the National Institute on Aging; the National Center for Research Resources; and NIH Roadmap for Medical Research under grants U01 AR45580, U01 AR45614, U01 AR45632, U01 AR45647, U01 AR45654, U01 AR45583, U01 AG18197, U01AG027810, and UL1 TR000128. The National Heart, Lung, and Blood Institute provides funding for the MrOS Sleep ancillary study Outcomes of Sleep Disorders in Older Men under the following grant numbers: R01 HL071194, R01HL070848, R01 HL070847, R01 HL070842, R01 HL070841, R01 HL070837, R01HL070838, and R01 HL070839. National Heart, Lung, and Blood Institute R21HL108226 and RO1 1 R01 HL 109493 funding supported this work.
Dr. Mehra has received NIH funding for which she has served as Principal Investigator (NHLBI RO1 1 R01 HL 109493, R21 HL108226); her institution has received positive airway pressure machines and equipment from Philips Respironics for use in NIH-funded research; she has received an honorarium from the American Academy of Sleep Medicine for speaking; she serves as the Associate Editor for the journal CHEST; and she has received royalties from UpToDate. Dr. Redline’s institution has received grant funding from ResMed, Inc., Philips Respironics, and ResMed Foundation; and equipment from them for use in NIH studies. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
Listen to this manuscript's audio summary by JACC: Clinical Electrophysiology Editor-in-Chief Dr. David J. Wilber.
- Abbreviations and Acronyms
- atrial fibrillation
- apnea-hypopnea index
- central apnea index
- confidence interval
- high-frequency power
- hazard ratio
- heart rate variability
- low-frequency power
- long-term variability
- obstructive apnea-hypopnea index
- premature atrial contractions
- sleep-disordered breathing
- short-term variability
- Received March 30, 2016.
- Revision received August 26, 2016.
- Accepted September 1, 2016.
- 2017 American College of Cardiology Foundation
- Peppard P.E.,
- Young T.,
- Barnet J.H.,
- Palta M.,
- Hagen E.W.,
- Hla K.M.
- Ghias M.,
- Scherlag B.J.,
- Lu Z.,
- et al.
- ↵Van Wagoner DR, Piccini JP, Albert CM, et al. Progress toward the prevention and treatment of atrial fibrillation: a summary of the Heart Rhythm Society Research Forum on the Treatment and Prevention of Atrial Fibrillation, Washington, DC, December 9–10, 2013. Heart Rhythm 2015;12:e5–29.
- ↵(1996) Heart rate variability: standards of measurement, physiological interpretation and clinical use. Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology. Circulation 93:1043–1065.
- Lippman N.,
- Stein K.M.,
- Lerman B.B.
- ↵Press, WH Rybicki GB. Fast algorithm for spectral analysis of unevenly sampled data. Astrophys J 1989;338:277–80.
- Akselrod S.,
- Gordon D.,
- Ubel F.A.,
- Shannon D.C.,
- Berger A.C.,
- Cohen R.J.
- Aviles R.J.,
- Martin D.O.,
- Apperson-Hansen C.,
- et al.
- Soliman E.Z.,
- Howard G.,
- Meschia J.F.,
- et al.
- Berry R.B.,
- Budhiraja R.,
- Gottlieb D.J.,
- et al.