Atrial signal amplitude predicts atrial high‐rate episodes in implantable cardioverter defibrillator patients: Insights from a large database of remote monitoring transmissions

Abstract Background Parameters measured during implantable cardioverter defibrillator (ICD) implant also depend on bioelectrical properties of the myocardium. We aimed to explore their potential association with clinical outcomes in patients with single/dual‐chamber ICD and cardiac resynchronization therapy defibrillator (CRT‐D). Methods In the framework of the Home Monitoring Expert Alliance, baseline electrical parameters for all implanted leads were compared by the occurrence of all‐cause mortality, adjudicated ventricular arrhythmia (VA), and atrial high‐rate episode lasting ≥24 hours (24 h AHRE). Results In a cohort of 2976 patients (58.1% ICD) with a median follow‐up of 25 months, event rates were 3.1/100 patient‐years for all‐cause mortality, 18.1/100 patient‐years for VA, and 9.3/100 patient‐years for 24 h AHRE. At univariate analysis, baseline shock impedance was consistently lower in groups with events than without, with a 40 Ω cutoff that better identified high‐risk patients. However, at multivariable analysis, the adjusted‐hazard ratios (HRs) lost statistical significance for any endpoint. Baseline atrial sensing amplitude during sinus rhythm was lower in patients with 24 h AHRE than in those without (2.45 [IQR: 1.65‐3.85] vs 3.51 [IQR: 2.37‐4.67] mV, P < .01). The adjusted HR for 24 h AHRE in patients with atrial sensing >1.5 mV vs those with values ≤1.5 mV was 0.52 (95% CI: 0.33‐0.83), P = .006. Conclusions Although lower baseline shock impedance was observed in patients with events, the association lost statistical significance at multivariable analysis. Conversely, low sinus rhythm atrial sensing (≤1.5 mV) measured with standard transvenous leads could identify subjects at high risk of atrial arrhythmia.


| INTRODUC TI ON
During implant of implantable cardioverter defibrillators (ICDs) and cardiac resynchronization therapy defibrillators (CRT-Ds), pacing threshold, impedance, and sensing amplitude are routinely assessed for all implanted leads. Their monitoring during follow-up with regular in-office visits or remote control is then used for the surveillance of integrity and functioning of leads.
Beyond technical aspects, several factors may influence these measurements, including properties of the myocardial tissue surrounding the lead electrodes. 1,2 Prior studies suggested an association between temporal changes in some of these parameters and clinical events, such as heart failure functional class changes or cardiac arrhythmias occurrence. [3][4][5] However, it is unknown whether their values at implant could have a clinical relevance as a systematic analysis of association between baseline measurements and clinical outcomes has never been performed.
The aim of the present study was to explore whether baseline electrical parameters routinely measured during implant have an association with long-term mortality or incidence of atrial and ventricular arrhythmias (VAs) in ICD/CRT-D recipients.

| ME THODS
The present analysis was performed in the framework of the Home Monitoring Expert Alliance (HMEA), an independent scientific project based on a nationwide repository of data generated by remote monitoring (RM) of cardiac implantable electronic devices (CIEDs) during ordinary medical practice. 6 A total of 41 Italian sites, listed in the Appendix, provided data for this analysis. All included patients provided written informed consent before RM activation.

| Objective and patient selection
We aimed to investigate whether baseline electrical parameters routinely measured during CIED implant could show an association with long-term mortality or incidence of atrial and VAs.
All ICD and CRT-D recipients registered in the HMEA database were selected for the present analysis. Devices were implanted from January 2007 to March 2017.

| Data collection and analysis endpoints
Baseline data were collected at the time of device implant. They included patient characteristics and electrical parameters routinely measured for any implanted lead during ICD/CRT-D implant procedures. Pacing impedance, capture threshold, and sensing amplitude were collected for atrial, right ventricle (RV), and left ventricle (LV) leads. Electrical parameters were usually obtained in bipolar configuration, expect for a minority (2.7%) of unipolar LV leads. In addition, highvoltage shock impedance, measured between distal coil and case, was reported for all RV leads. Atrial parameters were not included in the present analysis if the patient was in atrial fibrillation (AF) at implant. All CIEDs were manufactured by the same company (Biotronik) and follow-up data were automatically and daily generated by the RM system (Home Monitoring; Biotronik), which provided device diagnostics and intracardiac electrogram (IEGM) recordings of all atrial and ventricular arrhythmic episodes. Atrial high-rate episodes (AHREs) were recorded and transmitted based on a rate criterion set at 200 bpm (standard setting of devices). The atrial sensitivity adapted automatically on an ongoing basis to the measured amplitude of the atrial activity. The lowest sensing threshold that can be reached was 0.1 mV. VAs were automatically classified by the discrimination algorithms of the device. The median detection cutoff programed for the first VA zone was 158 (IQR: 150-171) beats/min with a counter of 28 For the AHRE analysis, the 24 hour duration threshold was used because it could be the sign of impaired atrial tissue and has recently been reported as the duration associated with an increased risk of ischemic stroke or systemic embolism. 7 Short-lasting atrial arrhythmias seem to have less clinical significance and an immediate anticoagulation in these patients is unlikely to result in reduction of the risk of stroke. 8 Only patients who had atrial diagnostics capability were selected for this endpoint.
In order to investigate whether some of the baseline electrical parameters could be a marker of endpoint occurrence, variables that had significant differences in the descriptive analysis were used to stratify event rates by value classes. The endpoints were then compared between the two subgroups defined using the value that maximized the difference in event rate as the cutoff value.

| Statistical analysis
We described the selected population by using all-cause death, VA,

| Baseline electrical parameters and allcause death
Some of the baseline electrical parameters were statistically different between survivors and deceased patients (

| Markers of arrhythmia occurrence and prognosis
At the descriptive analysis, the baseline shock impedance showed consistent differences for all study endpoints with lower median values in patients who experienced death, VA, and AHRE.  Table 4 reports the association of the adjusting covariates of the multivariate models.
The incidence of 24 h AHRE was significantly different according to baseline atrial signal amplitude measured in sinus rhythm (Table 5).

| D ISCUSS I ON
In the present analysis on about 3000 ICD and CRT-D patients, we found some associations between electrical parameters at implant and long-term clinical outcomes. Baseline shock impedance values were lower in patients with atrial and VAs and in those who died during follow-up. A cutoff of 40 Ω identified a subgroup with a particularly high incidence of events; however, the association was not significant if adjusted by other patients' characteristics.
Conversely, the higher incidence of atrial arrhythmias in patients with baseline atrial sensing in sinus rhythm ≤1.5 mV compared to >1.5 mV was statistically significant even after adjustment with patient characteristics. This parameter could be used as a potential marker of underlying atrial tissue disease, potentially identifying patients who may benefit from an intensive monitoring approach which can be provided by daily RM. Note: Event rates are expressed as events/100 patient-years (95% confidence interval).

TA B L E 3 Event rates of death, VA, and 24 h AHRE occurrence by baseline shock impedance subgroups
Only devices with atrial sensing capability were included in the 24 h AHRE analysis.
Abbreviations as listed in Table 1.

| Predictors of death and ventricular arrhythmias
Electrical data obtained during device implant can be influenced by several factors, as lead heart contact, lead position, and lead characteristics. However, bioelectrical properties can also be modified by other factors, including ischemia or the presence of fibrosis. In addition, shock impedance, which is calculated between the distal part of the lead (ventricular coil) and the ICD, depends on the conduction characteristics of the thorax including the whole heart and lungs.
Therefore, low values can be observed when electrical conduction is favored, as in the case of fluid overload. As a result, deceased and VA/AHRE patients showed lower baseline shock impedance in our study, probably as a consequence of increased lung congestion and more severe heart failure symptoms even at implant. However, we were not able to detect a significant association in multivariable models between baseline values and events occurrence. The association between decreased shock impedance and heart failure or VA was shown in several studies, but the temporal relationship between the events is still unclear. 5 We also found differences in the left ventricular pacing impedance, which is a near-field measurement, as calculated mainly in bipolar configuration. Patients who died during follow-up had lower baseline values than survivors. This result may reflect a higher percentage of patients with the LV lead located in an ischemic zone since infarct scar showed lower electrical impedance than the normal myocardium. 1,9 Finally, lower signal amplitudes for both right and LV leads were found in patients with death and VA events. This is not surprising as cytopenia and fibrosis are associated with lower signals and more advanced heart disease. 4,10 However, it should be noted that these parameters were statistically different only at the univariate analysis and the difference between groups was very small and with questionable clinical significance.

| Predictors of AHRE
The difference of atrial signal amplitude at implant between patients with and without AHRE later detected during follow-up was more striking.
It is well known that in the atria low signal amplitude is associated with the presence of scar, 9 AF recurrences, 11 and heart failure. 12,13 However, most data were obtained during atrial mapping for AF More intriguing is the relationship found between AHRE and shock impedance, confirming the association between atrial arrhythmias and severity of heart disease, 15 despite the difference, although statistically significant, was clinically quite negligible.

| Limitations
This study is an observational retrospective analysis suffering from all the known limitations of this design. Leads were placed according to clinical practice without specific recommendations and were not verified with fluoroscopy images, excluding the use of a variable lead location as adjusting covariate in our models. However, the large sample size of the database in terms of patients and sites is an important strength of this analysis tempering potential biases.
As the HMEA database is based on the Home Monitoring system, all devices included in the present analysis were made by Biotronik and this could have an impact on the detection algorithms of AHRE, signals, and impedance measurements.
Atrial high-rate episodes were not adjudicated potentially including far-field artifact and noise. However, the impact of the adjudication has been shown to be less relevant when using relatively long thresholds for diagnosis. The positive predictive value of AHRE increased to 98.2% when the threshold duration was prolonged to 24 h as in our analysis. 16 Finally, device programing was not uniform reflecting ordinary medical practice. As a relevant proportion of the devices included in our cohort were implanted before 2014 when more aggressive antitachycardia settings and shorter detections were largely used, the rate of VA may be higher as compared to contemporary cohorts.

ACK N OWLED G EM ENTS
The authors are grateful to all field engineers of BIOTRONIK Italia for their continuous technical support.