Subcutaneous implantable cardioverter-defibrillator noise following left ventricular assist device implantation

Abstract

Background

The incidence and impact of noise in a subcutaneous implantable cardioverter defibrillator (S-ICD) after left ventricular assist device (LVAD) implantation is not well established.

Methods

We performed a retrospective study of patients implanted with LVAD and with a pre-existing S-ICD between January 2005 and December 2020 at the three Mayo Clinic centers (Minnesota, Arizona, and Florida).

Results

Of the 908 LVAD patients, a pre-existing S-ICD was present in 9 patients (mean age 49.1 ± 13.7 years, 66.7% males), 100% with Boston Scientific third-generation EMBLEM MRI S-ICD, 11% with HeartMate II (HM II), 44% with HeartMate 3 (HM 3), and 44% with HeartWare (HW) LVAD. The incidence of noise from LVAD-related electromagnetic interference (EMI) was 33% and was only seen with HM 3 LVAD. Multiple measures attempted to resolve noise, including using alternative S-ICD sensing vector, adjusting S-ICD time zone, and increasing LVAD pump speed, were unsuccessful, necessitating S-ICD device therapies to be turned off permanently.

Conclusions

The incidence of LVAD-related S-ICD noise is high in patients with concomitant LVAD and S-ICD with significant impact on device function. As conservative management failed to resolve the EMI, the S-ICDs had to be programmed off to avoid inappropriate shocks. This study highlights the importance of awareness of LVAD-SICD device interference and the need to improve S-ICD detection algorithms to eliminate noise.

INTRODUCTION

Heart failure is associated with significant morbidity and mortality and its prevalence continues to rise over time.¹ From 2006 to 2014, advanced heart failure patients had mechanical circulatory support device implantation at a number exceeding 2000 per year.^2–4 Greater than 90% of these patients were implanted with a left ventricular assist device (LVAD).⁵ A majority of LVAD patients have a pre-existing implantable cardioverter defibrillator (ICD). Transvenous ICD (TV-ICD) implantation is associated with a significant risk of short term and long-term complications including complications from lead extraction and mortality.^6–9 The subcutaneous implantable cardioverter-defibrillator (S-ICD) is an entirely extra thoracic alternative to TV-ICD that received FDA approval in 2012 for prevention of sudden cardiac death in select patients.¹⁰ S-ICD use is recommended in patients meeting criteria for defibrillator therapy¹¹ who lack adequate vascular access, are at a high risk for infection, or who do not have a current or anticipated need for bradycardia pacing, anti-tachycardia pacing (ATP), or cardiac resynchronization therapy (CRT).¹² Electromagnetic interference (EMI) between LVAD and TV-ICD is an established phenomenon which can precipitate lead noise or telemetry interference, and inappropriate shocks.^13,14 The 2017 AHA/ACC/HRS Guideline for Management of Patients With Ventricular Arrhythmias and the Prevention of Sudden Cardiac Death recommend implantation of ICD in patients who are candidates for LVAD or OHT.¹² Therefore, the phenomenon of EMI in patients with LVAD and ICD is a significant clinical concern. Only small case series and isolated case reports are present in the literature regarding the phenomenon of EMI and sensing abnormalities in patients with concomitant LVAD and S-ICD. Therefore, we sought to study the incidence, clinical impact, and management of LVAD-related EMI in S-ICD patients at three tertiary care referral centers.

METHODS Study design and participants

We performed a retrospective analysis of patients implanted with LVAD at the three Mayo Clinic centers (Minnesota, Arizona, Florida) between January 1, 2005, and December 31, 2020. Patients with a pre-existing S-ICD at the time of LVAD implantation were identified and included in the study as detailed in Figure 1. The study protocol was approved by the Mayo Clinic Institutional Review Board.

FIGURE 1. Study enrollment algorithm.

Data collection

All patients were followed-up at the Heart Rhythm Device Clinic for at least 1 year after LVAD implantation. Two study personnel (B.K.K. and A.I.J.) reviewed patient charts and collected information on baseline demographic and clinical variables, including the etiology of cardiomyopathy, left ventricular ejection fraction (LVEF) and LV end diastolic diameter (LVEDD) at the time of LVAD implantation, LVAD indication and manufacturer, date of orthotopic heart transplantation (OHT), and mortality. Patient demographics and clinical characteristics were summarized with mean and standard deviation (SD) for continuous variables and frequency (%) for categorical variables.

ICD interrogation reports were reviewed to identify provider documentation of episodes of LVAD-related noise or EMI. When noise was reported, two study personnel (J.Z.L. and K.S.) independently reviewed the electrograms.

RESULTS

Out of the 908 patients implanted with LVAD between 2005 and 2020, a pre-existing S-ICD was present in 9 patients (mean age 49.1 ± 13.7 years, 66.7% males). All 9 patients were implanted with Boston Scientific third-generation EMBLEM MRI S-ICD. The LVAD types implanted were HeartMate II (HM II), HeartMate 3 (HM 3), and HeartWare (HW) in 11%, 44%, and 44% of the patients, respectively. The baseline and clinical characteristics of S-ICD patients with and without LVAD-related EMI were compared (Table 1). The incidence of LVAD-related noise in S-ICD was 33% (3 out of 9). All 3 patients with EMI had HM 3 LVAD, which represented 75% of all patients with LVAD and S-ICD. Patients with HM II and HW did not have EMI. A description of each of the 3 cases is outlined below and in Table 2. Multiple measures were unsuccessful in resolving the noise events, including utilizing other sensing vectors of the S-ICD, adjusting the time zone of the S-ICD, and increasing the LVAD pump speed, requiring S-ICD to be turned off permanently.

TABLE 1 Baseline demographics and clinical variables of patients with LVAD and S-ICD.

Characteristics	All patients (N = 9)	EMI (N = 3)	No EMI (N = 6)
Age (years)	49.1 ± 13.8	54.3 ± 20.3	46.5 ± 10.7
Males	6 (66.7%)	3 (100%)	3 (50%)
Cardiomyopathy
Ischemic	2 (22%)	0 (0%)	2 (33%)
Non-ischemic	7 (78%)	3 (100%)	4 (67%)
LVEF (%)	15.0 ± 7.2	9.3 ± 1.2	17.8 ± 7.4
LVEDD (mm)	67.9 ± 7.7	69.0 ± 10.8	67.3 ± 6.8
LVAD manufacturer
HeartMate II	1 (11%)	0 (0%)	1 (17%)
HeartMate 3	4 (44%)	3 (100%)	1 (17%)
HeartWare	4 (44%)	0 (0%)	4 (67%)
Orthotopic heart transplantation	5 (56%)	0 (0%)	5 (83%)
All-cause mortality	2 (22%)	0 (0%)	2 (33%)

Note: Continuous variables are in Mean ± SD; Categorical variables are in N (%).

Abbreviations: BTT, bridge to transplantation; DT, destination therapy; LVAD, left ventricular assist device; LVEDD, left ventricular end diastolic diameter; LVEF, left ventricular ejection fraction.

TABLE 2 Noise in patients with subcutaneous ICD after LVAD implantation.

No.	LVAD	S-ICD model	Noise	Management	Follow-upa (months)
1.	HM 3	A209 EMBLEM MRI	Noise in primary and secondary vectors and inappropriate sensing in alternate vector	ICD therapies off	3.4
2.	HM 3	A219 EMBLEM MRI	Noise in all three vectors	ICD therapies off	27.6
3.	HM 3	A219 EMBLEM MRI	Noise in all three vectors	ICD therapies off	23.7

Abbreviations: HM3, HeartMate 3; LVAD, left ventricular assist device; S-ICD, subcutaneous implantable cardioverter-defibrillator.

^aTotal follow-up duration is determined from date of LVAD implantation to the date of last follow-up in the heart rhythm clinic.

Case 1

A 64-year-old man with a history of end-stage heart failure LVEF of 10–15% and multiple recent hospitalizations for acute exacerbation of congestive heart failure was referred for LVAD evaluation. He was implanted with HM3 LVAD for destination therapy (DT). He was diagnosed with nonischemic dilated cardiomyopathy 7 years prior and had an S-ICD placed for primary prevention of sudden cardiac death 5 years prior to LVAD implantation after his EF dropped to 10%–15% (Figure 2). S-ICD interrogated 1 month prior to LVAD implantation showed that the device sensed appropriately in all three vectors (primary, secondary, and alternate). S-ICD interrogation immediately after LVAD implantation revealed noise in the primary and secondary vectors and intermittent undersensing in the alternate vector (Figure 3). The noise and inadequate sensing persisted on 2-week follow-up in the device clinic. An attempt was made to resolve these complications by increasing the LVAD pump speed and changing the device time zone from Eastern Standard Time (EST) to Greenwich Mean Time (GMT) to reduce the built-in notch filter from 60-Hz to 50-Hz. At 1-month follow-up, these strategies were found to be unsuccessful as the noise persisted. As a last resort, ICD therapy was programmed off. At 3-month follow-up, telemetry monitoring showed only infrequent episodes of non-sustained ventricular tachycardia (NSVT), with the longest episode being four beats long. After shared decision-making with the patient, TV-ICD was not implanted in exchange for S-ICD, and S-ICD therapy remained permanently off.

FIGURE 2. Chest radiograph of a patient with concomitant HeartMate 3 left ventricular assist device (LVAD) and subcutaneous implantable cardioverter-defibrillator (S-ICD). The pulse generator of the S-ICD lies in proximity with the LVAD pump. The three sensing vectors shown are the primary vector (ring electrode to pulse generator), secondary vector (tip electrode to pulse generator) and alternate vector (tip electrode to ring electrode).

FIGURE 3. Electrogram for patient number 1 showing noise in primary and secondary vectors with inappropriate sensing in the alternate vector after LVAD implantation. The device classifies a signal as noise when it exceeds the range of sensing amplifier. There is a noisy baseline, and the device marks the biggest signal (R wave in this case) with the marker “N”.

Case 2

A 31-year-old man with a history of amphetamine use developed non-ischemic cardiomyopathy with LVEF of 20%. He ceased amphetamine use after the diagnosis of heart failure and remained abstinent throughout a period of 4 years, after which he had an S-ICD implanted. S-ICD interrogation at the time of initial implantation revealed normal device function. Despite abstinence from drugs and adherence to guideline directed medical therapy (GDMT) for heart failure, his LVEF remained less than 30%. After recurrent hospitalizations for heart failure, he eventually had a HM3 LVAD implanted as a bridge to transplant therapy (BTT) 15 months after his S-ICD implantation. All three vectors of the S-ICD (primary, secondary, and alternate) sensed appropriately prior to LVAD implantation. When interrogated 1 day post LVAD implantation, the S-ICD sensed noise in all three vectors. Over the following months, due to psychosocial issues and inability to show reliable compliance with therapy and follow-up, he was no longer a candidate for orthotopic heart transplant (OHT). Noise in all vectors (Figure 4) has persisted over a period of 27 months of device clinic follow-up, necessitating ICD therapy to be programmed off permanently.

FIGURE 4. Electrogram for patient number 2 showing noise detections on the marker channel of the S-ICD from HeartMate 3 LVAD post-LVAD implantation.

Case 3

A 68-year-old man with a history of nonischemic cardiomyopathy with LVEF of 10% had a S-ICD implanted for primary prevention of sudden cardiac death. S-ICD sensed appropriately in all three vectors and there were no inappropriate or appropriate shocks. Two years later, due to refractory New York Heart Association (NYHA) class IV heart failure symptoms, he underwent LVAD implantation as DT with a HM3 device. Upon interrogation in the immediate postoperative period, noise was found to be present in all three vectors, necessitating S-ICD therapy to be programmed off. After 2 months, the EMI was still present during a device clinic visit (Figure 5). Noise persisted during subsequent follow-ups, so S-ICD therapy was programmed permanently off. As the patient only had occasional episodes of NSVT noted on ambulatory monitoring during follow-up, TV-ICD was not implanted.

FIGURE 5. Electrogram for patient number 3 showing noise in primary and secondary vectors and inappropriate sensing in alternate vector post-LVAD implantation.

DISCUSSION

Our retrospective analysis of patients with concomitant S-ICD and LVAD had several important findings. First, S-ICD noise due to interference from LVAD was a common occurrence (33% of patients). Second, LVAD-related EMI is common in patients with HM 3 (75% of patients) and was not seen in patients with HM II or HW. Third, LVAD-related S-ICD noise has significant clinical impact as conservative methods did not resolve EMI, resulting in the need to permanently turn off ICD therapy.

There is conflicting evidence regarding the benefit of ICD implantation in patients with existing LVAD. As LVADs provide hemodynamic support in ventricular arrhythmias, the utility of ICDs in these patients has been debated. In a subgroup analysis of patients in the United Network for Organ Sharing registry listed for heart transplant from 1999 to 2014, 9478 patients were found to have LVAD, of which 6529 had an ICD. The presence of ICD was associated with an adjusted 19% relative reduction in mortality (HR: 0.81; 95% CI: 0.70–0.94). This study included many patients with pulsatile flow LVADs. In a 2019 retrospective of patients with continuous flow LVADs a subgroup analysis of 99 patients showed that ICD implantation after LVAD was not associated with mortality benefit. However, these two studies have not analyzed patients with S-ICD. Much of the morbidity in these studies was associated with pocket or transvenous lead complications, which would not be present in patients with S-ICD.

Although LVADs can provide hemodynamic support during ventricular arrhythmias, ventricular arrhythmias over time can lead to right ventricular dysfunction which can compromise LVAD function.^15,16 Therefore, detection of ventricular arrhythmias by S-ICDs and subsequent S-ICD therapy can theoretically aid clinicians in adjusting medical therapy and preventing RV deterioration,¹⁵ although this has not been analyzed in large studies. The most recent AHA/ACC/HRS Guideline for Management of Patients with Ventricular Arrhythmias and the Prevention of Sudden Cardiac Death issued a Class IIa recommendation for ICD implantation in non-hospitalized patients with NYHA class IV symptoms who are candidates for cardiac transplantation or an LVAD if meaningful survival of greater than 1 year is expected.¹² Consequently, ICD therapy continues to have a role in patients with existing LVAD.

S-ICD pulse generator and vectors

S-ICDs have a pulse generator and two sensing electrodes, proximal and distal. Rhythm detection can be performed by any of the three sensing vectors: primary, secondary, or alternate.¹⁷ The primary sensing vector is from the proximal electrode to the pulse generator. The secondary sensing vector is from the distal electrode to the pulse generator. The alternate sensing vector is from the distal electrode to the proximal electrode (Figure 2).

LVAD-related S-ICD noise

The use of an entirely subcutaneous ICD was first reported in 2010.¹⁸ The first case of concomitant S-ICD and LVAD was reported in 2013 by Saeed et al. In most cases reports in the literature, noise was detected in the primary and secondary vectors of the S-ICD but not in the alternate vector.^19–23 The S-ICD device tracks directional changes in incoming signals within the refractory period or sensing window after a signal. The device classifies a signal as noise when it exceeds the range of the sensing amplifier and labels the most prominent signal in the refractory window (in most cases the R wave) with the marker “N” (Figure 3).¹⁹ In 2018, Afzal et al. reported a multicenter cohort of 20 patients with concomitant S-ICD and LVAD of which only one patient out of 20, sensed EMI in all three vectors.²¹ In contrast, all three patients with EMI in our study had noise in all three vectors. The primary and secondary vectors include the pulse generator of the S-ICD which is spatially close to the LVAD (Figure 2). The ring and tip electrodes of the alternate vector are at a further distance from the LVAD, which may explain why EMI was not seen with this vector but undersensing is sometimes detected.^19,20,24

Similar to our study, Gupta et al (2015) and Raman et al (2016) reported that no EMI occurred between HM II LVAD and S-ICD.^24,25 The proposed mechanism of noise is the electromagnetic field generated by the LVAD. The fundamental frequencies responsible for EMI are directly proportional to the LVAD pump speed. The HM II LVAD is an axial-flow pump that operates at a higher rotational speed of 8800 to 10 000 RPM. The higher rotational speed of HM II is less likely to cause interference. On the other hand, HW operates at a frequency range of 1800 to 3200 RPM, which is within the detectable frequency range by the S-ICD and is classified by signal detection/certification algorithms as noise.¹⁹ Additionally, the HM 3 LVAD has a different structure than HM 2. The HM 3 is a centrifugal continuous-flow LVAD which uses magnetic levitation. It also has a fixed pulse which is asynchronous with the native heartbeat. On the other hand, the HM 2 is a continuous flow pump which uses mechanical bearings associated with greater friction. These structural differences may possibly lead to detected noise. Saeed et al reported 1 of 2 patients, Gupta et al reported 2 of 3 patients, and Ahmed et al reported one patient with EMI in S-ICD after HW LVAD implantation. In our study, none of the four HW LVAD patients had EMI.

In a case series by Black-Maier et al. in 2020, four patients with concomitant S-ICD and LVAD were identified.²² Two of the patients with EMI received inappropriate shocks from EMI. Three of the patients in the case series by Black-Maier et al had HM 3, of whom all had clinically significant EMI, with one patient necessitating therapies to be turned off. Similarly, three HM 3 patients had EMI necessitating tachycardia therapies to be turned off. In their systematic review, no cases of EMI were found in the 11 patients with HM II, which is consistent with our study finding and the low rate of clinical events in HM II patients in Afzal et al's study in 2018.²¹ This finding is important considering the potential increase in concomitant use of LVAD and S-ICD in the future while avoiding device interference.

Unlike in TV-ICD, noise in S-ICD rarely leads to inappropriate therapies and in fact appropriate therapy may not be delivered if a true arrhythmia occurs. This is because the waveform appraisal algorithm ensures that noise detections are not included in heart rate calculations making the system unable to detect and treat arrhythmias when noise is present.²⁶ In our study no inappropriate shocks were detected, unlike the study by Black-Maier et al.

S-ICDs detect signals in a frequency range of 0–60 Hz.¹⁹ Ahmed et al. in 2018 demonstrated a dominant frequency of 46.67 Hz in the primary and secondary vectors which correlated well with the operational speed of 2800 RPM of the HW LVAD. In their study, increasing the HW LVAD speed from 2800 to 3000 RPM and reprogramming the S-ICD time zone from EST to GMT to change the default notch filter from 60 to 50 Hz eliminated EMI in supine position but not in the upright position. The S-ICD was turned off given no documented history of ventricular arrhythmias or ICD therapies. Changing the S-ICD time zone can be helpful in some instances as during initial implantation, S-ICD notch filter is automatically configured as either 60 Hz in the United States (U.S.) or 50 Hz (in Europe) based on the time zone selected. The notch filter removes powerline frequencies and the settings are different in the U.S. and Europe due to variation in anticipated noise from the prominent electrical grid within the region.^19,26

In Afzal et al's study, one out of nine patients with HM II received an inappropriate shock, which was due to P-wave oversensing. Two out of nine patients with HW experienced EMI from LVAD, of which one patient had ICD therapies permanently off due to persistence of EMI in all three vectors and a TV-ICD was subsequently implanted. The second patient with HW related EMI had the device programmed to the alternate vector which sensed appropriately. In their study, data was limited regarding EMI with HM 3 LVAD, as only two patients had this LVAD type.

An important finding in the Black-Maier et al study was that over a period of 1–3 months, two of the three HM 3 patients with EMI did well clinically and were able to use the alternate vector for sensing, which allowed ICD therapies to be re-enabled, eliminating the need for device revision or extraction. This was not apparent in our case series, as the three of four patients with HM 3 had S-ICDs turned off permanently because noise persisted in all three vectors, including the alternate vector. The HM 3 patient who did not experience EMI received OHT within 1 year, while the other three patients did not. This indicates that the noise in the HM 3 patients was a delayed phenomenon.

There have been multiple case reports of patients with HM 3 with clinical events necessitating S-ICD therapies to be turned off permanently, with two of these patients requiring S-ICD removal and TV-ICD implantation.^27–29 The relatively low number of HM 3 patients in the study by Afzal et al may explain the relatively lower occurrence of noise issues. There appears to be a developing trend of significant noise in patients with concomitant S-ICD and HM 3, and it would be prudent to closely monitor this population. These findings are also significant considering recent LVAD implantation trends. Due to the enhanced safety profile of HM 3 in the MOMENTUM-3 Trial, the number of HM 3 implants has significantly increased, while HM II implants have significantly declined.³⁰

Management

Multiple conservative approaches have been suggested to overcome EMI and sensing abnormalities. One approach is to use the secondary or alternate sensing vector. The S-ICD is programmed to automatically select the best vector for sensing. In some cases, manually changing the sensing vector to alternate vector can resolve LVAD-related noise, although as evident in our study, this approach may be unsuccessful. Another strategy is to increase the LVAD pump speed. Studies have shown that sensing abnormalities are more common in LVADs with lower RPM,³¹ which is likely why EMI has not been found with HM II. In our study, changing the notch filter from 60 to 50 Hz was not successful in reducing noise. Finally, watchful waiting can be beneficial, as sensing and noise issues in some patients will resolve spontaneously with time with improvements in postoperative inflammation and edema, although this was ineffective in our patient cases.

In this case series, in the patients in whom noise was present, when SICD therapy was turned off, there was no associated significant morbidity. The electrophysiologists supervising care of the patients did not deem it necessary to implant a TV-ICD over a follow up maximum of 3, 24, and 27 months respectively. This was because TV-ICDs carry unique risks of morbidity such as lead or pocket complications. However, it is possible over a longer follow up period, ventricular tachycardia can cause RV deterioration, impairing the ability of the LVAD.^15,16 Therefore, patients who are unable to receive OHT may theoretically benefit from implantation of TV-ICD.

Clinical perspective

It is important to ensure that patients who have had their S-ICD therapies programmed off continue to have follow-up in a device clinic to evaluate if noise persists months after LVAD implantation. The conservative measures described here are a logical first step in troubleshooting sensing and noise issues in patients with LVAD and S-ICD, as they have few if any drawbacks. However, it is also important to educate patients and providers that these measures may be inadequate, especially for HM 3 LVADs.

LIMITATIONS

This study is an observational, non-randomized analysis in a small sample of LVAD and S-ICD patients and is subject to confounding and selection bias. Consequently, it is unclear if continuing defibrillator function will be beneficial for post-LVAD patients. Ideally, a randomized controlled trial in this patient population comparing outcomes in patients with shock function turned OFF and patients with shock function kept ON would be necessary to provide guidance for optimal management.

CONCLUSIONS

We aim to highlight the issue of noise in patients with LVAD and S-ICD, to serve as a discussion point for providers faced with S-ICD management decisions post-LVAD implantation. LVAD-related noise secondary to EMI is a relatively common occurrence after LVAD implantation in patients with S-ICD, and its consequences significantly impact patients, especially those with HM 3 LVAD. Multiple measures attempted to resolve noise, including using alternative S-ICD sensing vector, adjusting S-ICD time zone, and increasing LVAD pump speed, were unsuccessful in this study, necessitating S-ICD device therapies to be turned off permanently. While turning off therapies in some patients may be appropriate, others may require replacement of the S-ICD with a TV-ICD, especially if the patient requires therapies for ventricular tachyarrhythmias. The S-ICD detection algorithm may require additional changes in bandpass filtering for appropriate certification and overcoming noise detections.

FUNDING INFORMATION

The authors received no funding from any sources.

CONFLICT OF INTEREST STATEMENT

Authors declare no conflicts of interest for this article.

ETHICS APPROVAL

The study was approved by Mayo Clinic Institutional Review Board.

PATIENT CONSENT STATEMENT

Not applicable.

CLINICAL TRIAL REGISTRATION

Not applicable.