Continuous monitoring of sleep‐disordered breathing with pacemakers: Indexes for risk stratification of atrial fibrillation and risk of stroke

Abstract Background Sleep apnea (SA) is a risk factor for atrial fibrillation (AF). Advanced pacemakers are now able to calculate indexes of SA severity. Hypothesis We investigated the changes in pacemaker‐measured indexes of SA, we assessed their predictive value for AF occurrence and the associated risk of stroke and death at long‐term. Methods We enrolled 439 recipients of a pacemaker endowed with an algorithm for the calculation of a Respiratory Disturbance Index (RDI). The RDI variability was measured over the first 12 months after implantation, as well as its potential association with the occurrence of AF, defined as device‐detected cumulative AF burden ≥6 hoursours in a day. Results The individual RDI mean was 30 ± 18 episodes/h, and the RDI maximum was 59 ± 21 episodes/h. RDI ≥30 episodes/h was detected in 351 (80%) patients during at least one night. The proportion of nights with RDI ≥30 episodes/h was 14% (2%‐36%). AF ≥6 hours was detected in 129 (29%) patients during the first 12 months. The risk of AF was higher in patients with RDI maximum ≥63 episodes/h (HR:1.74; 95%CI: 1.22‐2.48; P = .001) and with RDI mean ≥ 46 episodes/h (HR:1.63; 95%CI: 1.03‐2.57; P = .014). The risk of all‐cause death or stroke was higher in patients with AF burden ≥6 hours (HR:1.75; 95%CI: 1.06‐2.86; P = .016). Moreover, among patients with no previous history of AF the risk was higher in those with RDI maximum ≥63 episodes/h (HR:1.96; 95%CI: 1.06‐3.63; P = .031). Conclusions Pacemaker‐detected SA showed a considerable variability during follow‐up. We confirmed the association between RDI and higher risk of AF, and we observed an association between higher RDI maximum and all‐cause death or stroke among patients with no previous history of AF.


| INTRODUCTION
Sleep apnea (SA) is a well-known risk factor associated with increased incidence of cardiovascular morbidity, including atrial fibrillation (AF), 1 hypertension, 2 heart failure, and coronary artery disease. 3 SA has also been associated with a 3-fold increase in the risk of all-cause mortality at long term. 4 Moreover, in patients with sleep-disordered breathing, a doubled to tripled risk of stroke has been observed independently on other risk factors, [5][6][7] as well as an increased risk of all-cause mortality. 8 This association was stronger in males, 9 and was observed also in patients with AF already treated with oral anticoagulants. 10 SA has a prevalence of about 60% in patients implanted with pacemakers according to conventional indications. 11 Recently, automated pacemaker algorithms have been developed to calculate a Respiratory Disturbance Index (RDI) that well correlates with the results of standard polysomnography, 12,13 and that showed a good performance in identifying patients at higher risk of new-onset AF over the mid-term. 14,15 The availability of these algorithms offers the opportunity to monitor sleep-disordered breathing in the long run.
The aims of this study were to investigate the changes in sleepdisordered breathing severity, to calculate pacemaker-derived indexes of SA severity, to assess their predictive value for AF occurrence over 1-year follow-up and the associated risk of stroke and death at long-term.   12 SA was defined as severe if the pacemaker-measured RDI was ≥30 episodes/h. 12,17 The night-tonight variability of RDI values was investigated by measuring the following indexes in each patient: the average RDI value (RDI mean), the maximum value (RDI max), the proportion of nights with RDI ≥30 episodes/h (RDI burden). The RDI variability was measured over the first 12 months after implantation, as well as its potential association with the occurrence of AF in each patient.
The incidence and duration of AF were derived from device data, which comprise the total time spent by the patient in AF on each day of the follow-up period. Patients were considered to have experienced AF episodes if the device detected a cumulative AF duration greater than or equal to 6 hours in a day, in agreement with previous studies. 18,19 The patient outcome was assessed over the entire follow-up by measuring the combined endpoint of allcause death or stroke. Stroke was defined as rapidly developing signs of focal (or global) disturbance of cerebral function lasting more than 24 hours with no apparent nonvascular cause. 20 The diagnosis of ischemic stroke was established by a neurologist after standard diagnostic tests including brain imaging. The prognostic value of pacemaker-detected sleep-disordered breathing was tested by measuring the association between RDI parameters and the combined endpoint.

| Statistical analysis
Continuous data were expressed as mean ± SD or median and interquartile range (25th-75th percentile). Categorical data were expressed as percentages. Event rates were summarized by constructing Kaplan-Meier curves, and the distributions of the groups were compared by means of a log-rank test. Cox proportional hazards models were used to determine the association between covariates and the occurrence of AF during the first 12 months after implantation and to estimate the hazard ratios (HRs) and the 95% confidence intervals (CIs) of an AF event. All variables associated to a P value <.05 on univariate analysis were entered into the multivariate regression analysis. A P value <.05 was considered significant for all tests. A receiver operating characteristic (ROC) curve analysis was conducted to assess the performance of the RDI measures as predictors for AF, and we regarded the value resulting in the maximum product of sensitivity and specificity on the curve as the optimal cutoff. All statistical analyses were performed by means of the STATISTICA software, version 7.1 (StatSoft, Inc.).

| Study population
From October 2013 to March 2019, a total of 439 consecutive patients with a standard indication for permanent pacing underwent dual-chamber pacemaker implantation in our center. Table 1 shows baseline clinical variables. The pacemaker was implanted for atrioventricular block in 173 cases (40%), sinus node disease in 229 (52%), and carotid sinus syndrome in 37 (8%). One hundred sixty-three (37%) patients presented with a history of AF. Oral anticoagulant therapy was used by 134 (31%) patients on hospital discharge and by 166 (38%) at the time of the last follow-up visit.

| Sleep-disordered breathing during follow-up
During the 12-months post-implantation period, the individual average RDI value (RDI mean) was 30 ± 18 episodes/h, and the maximum value (RDI max) was 59 ± 21 episodes/h in the study population. The individual mean night-to-night RDI coefficient of variation was 34 ± 17%, with an absolute SD of 11 ± 6 episodes/h for each patient. RDI ≥30 episodes/h was detected in 351 (80%) patients during at least one night. The individual median proportion of nights with RDI ≥30 episodes/h (RDI burden) was 14% (25th -75th percentile: 2% -36%). The proportion of patients achieving RDI maximum values ≥30episodes/h slightly increased from the first observation after implantation to the 12-month visit ( Figure 1). Despite a wide variability, the assessment of RDI mean value and RDI burden value seemed not affected by the length of the observation period (Figure 1). Previous studies have shown that SA, as detected by conventional polysomnography, was more prevalent in patients with clinical AF. 1,23 Multiple mechanisms have been proposed to explain the link between SA and AF, such as the apnea-induced hypoxemia, 24 the diastolic dysfunction induced by the cardiac wall stress generated by the attempted inspirations, 25 the SA-induced systemic inflammation, 26 the autonomic imbalance that has been seen to occur during SA. 27 Atrial fibrillation is common in patients with pacemakers, 28 and devices frequently detect silent episodes in patients without a clinical history of AF. 29 Our findings that severe SA was independently predictive of AF occurrence extends to the specific field of device-detected subclinical AF, and confirm previous smaller studies. 14,30,31 Our results suggest that device-based RDI may be helpful in stratifying pacemaker patients for AF risk, by reviewing these automated diagnostic data during routine in-office or remote interrogation. Indeed, the risk of AF was almost doubled in patients with SA diagnosed by the device during follow-up.

T A B L E 1 Demographics and baseline clinical parameters
The optimal RDI values for the prediction of AF with pacemaker, that is, RDI max ≥63 episodes/h and RDI mean ≥ 46 episodes/h, were significantly higher than the value of apnea-hypopnea index (≥30 episodes/h) commonly adopted to diagnose severe SA with standard polysomnography. 12,17 This result is consistent with previous findings on the agreement between AHI and RDI values recorded during the sleep-study night. 32 Indeed, the RDI value was shown to be affected by a positive bias of 11 episodes/h. In a previous pacemaker study, 31 the nights with the highest pacemaker-detected SA severity seemed to confer an increased risk of having at least 1 hour of AF the same day compared to nights with the lowest SA severity. This seemed to suggest that a measure like the proportion of nights with high RDI (SDB burden), rather than a categorical diagnosis of SA per se, may be a more useful metric to determine SA severity in the management of concomitant AF. However, in our analysis RDI max and RDI mean were associated with AF occurrence during follow-up with comparable predictive power, while RDI burden seemed a weaker predictor.
The capability of implanted cardiac devices to monitor the atrial rhythm is an opportunity to stratify patients for the risk of ischemic stroke, and may constitute an effective tool for ensuring correct antithrombotic treatment. 19 It has previously been found that a maximum daily AF burden of 6 hours implies a 17% increase in the risk of stroke. 19 The present results seem to confirm this finding, indeed in our patients with pacemaker-diagnosed AF (daily AF burden ≥6 hours) the risk of allcause death or stroke was higher during a median follow-up of 32 months. Interestingly, we also observed an association between higher  33,34 By using this threshold of AF burden we were therefore not able to distinguish between patients with shorter AF duration and those with no AF. In agreement with previous analyses, 18,19,34 the 6 hours cutoff was chosen as an approximation of the 5.5 hours value identified in one of the studies evaluating the association between pacemaker-detected AF and thromboembolic risk. 33,34 In addition, an ECG analysis was not performed to include specific variables (eg, P-wave duration) in our analysis of predictors of AF occurrence.

| CONCLUSIONS
In the present study, pacemaker-detected SA showed a considerable variability during follow-up, suggesting that continuous monitoring might be superior to a single overnight sleep study for classifying SA severity. We confirmed the association between device-based RDI and higher risk of AF, and we also observed an association between higher RDI max values and the combined end-point of all-cause death or stroke among patients with no previous history of AF.

DATA AVAILABILITY STATEMENT
The dataset analyzed during the current study is available from the corresponding author on reasonable request.