Author + information
- Received June 12, 2018
- Revision received August 7, 2018
- Accepted August 17, 2018
- Published online January 21, 2019.
- Robert D. Anderson, MBBSa,
- Geoffrey Lee, MBChB, PhDa,
- Sohaib Virk, BMed/MDb,
- Richard G. Bennett, MBChBc,
- Christopher S. Hayward, BMedSc, MDd,
- Kavitha Muthiah, MBChB, PhDd,
- Jonathan Kalman, MBBS, PhDa and
- Saurabh Kumar, BSc(Med)/MBBS, PhDb,e,∗ ()
- aDepartment of Cardiology, Royal Melbourne Hospital, Faculty of Medicine, Dentistry, and Health Science, University of Melbourne, Melbourne, Australia
- bDepartment of Cardiology, Westmead Hospital, Westmead, Australia
- cBristol Heart Institute, Bristol Royal Infirmary, Bristol, United Kingdom
- dHeart Failure and Transplant Unit, Department of Cardiology, St. Vincent’s Hospital, Darlinghurst, Australia
- eDepartment of Cardiology, Westmead Applied Research Centre, University of Sydney, Westmead, Australia
- ↵∗Address for correspondence:
Assoc. Prof. Saurabh Kumar, Department of Cardiology, Westmead Hospital, Westmead Applied Research Centre, University of Sydney, Darcy Road, Westmead, New South Wales 2145, Australia.
Objectives This is a systematic review summarizing the procedural characteristics and outcomes of ventricular assist device (VAD)–related ventricular tachycardia (VT) ablation.
Background Drug-refractory VT refractory commonly develops post-VAD implantation. Procedural and outcome data come from small series or case reports.
Methods An electronic search was performed using major databases. Primary outcomes were VT recurrence, mortality, and cardiac transplantation. Secondary endpoints were acute procedural success and procedural complications.
Results Eighteen studies were included, with a total of 110 patients (mean age 59.6 ± 11 years, 89% men; VT storm 34%). Scar-related re-entry was the predominant mechanism of VT (90.3%) and cannula-related VT in 19.3% cases. Electroanatomical mapping interference occurred in 1.8% of cases; there were no reports of catheter entrapment. Noninducibility of clinical VT was achieved in 77.9%; procedural complications occurred in 9.4%. At a mean follow-up of 263.5 ± 267.0 days, VT recurred in 43.6%, 23.4% underwent cardiac transplant, and 48.1% died. There were no procedural-related deaths and no death was directly related to ventricular arrhythmia. In follow-up, there was a significant reduction in implantable cardioverter-defibrillator therapies or shocks (57.1% vs. 23.8%). Ablation allowed VT storm termination in 90% of patients.
Conclusions VAD-related VT is predominantly related to pre-existing intrinsic myocardial scar rather than inflow cannula site insertion. Catheter ablation is a reasonable treatment strategy, albeit with expectedly high rate of recurrence, transplantation, and mortality related to severe underlying disease.
- catheter ablation
- mechanical support
- radiofrequency ablation
- ventricular assist device
- ventricular tachycardia
Ventricular assist devices (VADs) are increasingly being used either as a bridge to cardiac transplantation, or as destination therapy for nontransplant candidates. Ventricular arrhythmias (VAs) are common in patients post-left VAD (LVAD) implantation and are associated with considerable morbidity from recurrent implantable cardioverter-defibrillator (ICD) shocks and progressive failure of the unsupported right ventricle, and portend an increased risk of mortality (1–3).
Antiarrhythmic drugs and ICDs are used as first-line treatment for VAD-related ventricular tachycardia (VT), but the arrhythmia may become refractory. Such patients have few options other than catheter ablation. Although catheter ablation reportedly plays a critical role in the control of VT in VAD patients, data come from a limited number of small case series and case reports. There are technical and procedural challenges perceived to be unique to VAD-related VT ablation such as catheter entrapment, perceived difficulty with catheter maneuverability within decompressed chambers, and electroanatomic mapping system (EAM) interference. However, the extent to which these challenges impact on procedural outcomes has not been well quantified, and much conflicting data exists. A summation of published data on the characteristics and origin of VT, procedural approaches, challenges, and outcomes in follow-up is needed to help contextualize the role of VT ablation in VAD patients. In this study, we performed a systematic review of published studies assessing the arrhythmia characteristics as well as clinical and procedural outcomes in this specialized group.
We searched PubMed, Embase, and the Cochrane Central Register of Clinical Trials (Cochrane Library, issue 3, March 2018) databases to identify studies that assessed the use of catheter ablation for VT after LVAD procedures. We used the terms (Ventricular tachycardia OR VT OR tachyarrhythmias) AND (ablation OR radiofrequency ablation OR RFA) AND (Ventricular assist device OR LVAD OR RVAD OR BIVAD or mechanical support). No language restriction was applied. The reference list of all identified articles was also reviewed for relevant publications fitting the eligibility criteria. No librarian assisted with the search process.
Studies with the following characteristics were considered eligible: 1) included patients with surgically implanted durable left ventricular assist device (LVAD), right ventricular assist device (RVAD), or biventricular assist device, and studies using percutaneous assist devices or extracorporeal membrane oxygenation circulatory support were excluded; and 2) VT that developed following implantation of a VAD, treated with catheter ablation. Reviews, abstracts, editorials, expert opinions and conference presentations were excluded from analysis.
Data extractions and quality appraisal
Two investigators (R.D.A. and S.K.) independently screened all titles and abstracts and manually searched the full-text versions of all relevant studies that fulfilled the inclusion criteria. References of the retrieved articles were independently reviewed for full identification of potentially relevant studies. Disagreements were resolved arbitrarily with consensus reached following careful examination and discussion. We extracted characteristics from each study including methodology, baseline patient demographic data, LVAD indication, type and planned use, VT characteristics, technique and success of ablation, complications, recurrence, cardiac transplantation, mortality, and 12-month follow-up.
The primary endpoints were VT recurrence, mortality, and transplantation. Additional primary endpoints were control of VT storm with catheter ablation and impact on defibrillator therapies pre-ablation versus post-ablation. We also evaluated, where available, characteristics and exit sites of spontaneous and inducible VTs, location of scar, mechanism of VT (scar-mediated, bundle branch re-entry, or focal or micro-re-entry), and acute procedural outcomes and complications. Acute procedural outcomes were reported as noninducibility of the clinical VT and, separately, noninducibility of any VT.
Acute procedural success, incidence of procedural complications, and primary and secondary endpoints were pooled using DerSimonian-Laird random-effects models. Summary estimates were reported with 95% confidence intervals (CIs). The I2 statistic was used to estimate the percentage of total variation across studies due to heterogeneity, with values exceeding 50% indicative of considerable heterogeneity. Statistical analysis was conducted with Comprehensive Meta-Analysis v3.3 (Biostat, Englewood, New Jersey).
A total of 124 studies were identified using the specified search criteria. After a detailed evaluation of these studies, 18 relevant studies that incorporated 110 patients were included (8 case series and 10 case reports). There were no prospective studies identified. One patient was excluded from a case series, as the VT developed before LVAD implantation (4). A flow algorithm is shown in Figure 1.
Baseline demographics and LVAD characteristics
Table 1 summarizes patient demographics. Among the cohort, the mean age was 59.6 ± 11.0 years (89.1% men) and the etiology was ischemic cardiomyopathy in 59.1%. The majority of patients (89.7%) had an ICD implanted before ablation. A history of VA before VAD implant was present in 71.9%; 14.5% of patients had undergone previous VT ablation. VT storm was an indication for ablation in 33.6% of patients.
LVAD characteristics were reported in 88 patients and are summarized in Table 2. The majority of patients (98.9%) were implanted with a continuous flow device (81.8% with HeartMate II [St. Jude Medical, St. Paul, Minnesota] and 9.1% had HeartWare HVAD device [HeartWare, Framingham, Massachusetts]). LVADs were planned as bridge therapy in 46.6% of patients and destination therapy in 53.4% of patients unsuitable for cardiac transplantation. In the studies that reported time from VAD to the development of VT, the median time was 60 (range: 0 to 493) days.
Catheter access to the LV was predominantly via a transseptal (TS) puncture (60.2%), followed by retrograde aortic (RA) access (36.4%) and epicardial access (2.3%). Dual TS and RA access was only used in 1.1% (Figure 2). LV electroanatomical mapping was recorded in 86 of 110 (78%) cases. VT induction was attempted in all cases with most common stimulation protocol used was RV stimulation down to 3 extrastimuli. Activation or entrainment was used in 61.2% of patients, followed by substrate mapping in 20.4% and pace mapping in 2.1%. A combination of these techniques was used in 16.3% of patients (Figure 2). All patients had scar identified, with 36.1% of cases in the septum, 33.7% in the basal region, 32.5% in the inferior wall, 32.5% in the anterior wall, and 26.5% in the lateral wall. Premature ventricular complex (PVC)–triggered VT or ventricular fibrillation was identified in 3 (2.7%) cases. Cases in which the 12-lead electrocardiogram (ECG) VT morphology was available were correlated to scar distribution, VT mechanism, and successful ablation site (based on detailed activation or entrainment or pace mapping in all cases) (Table 3). Notably, the 12-lead ECG morphology correlated with the VT exit site in only 5 of 11 (45%) cases. Twelve-lead ECG morphology suggestive of cannula-related VT correlated with site of VT exit in 4 of 6 (67%) cases.
VT was induced and tolerated in 86 (78.2%) patients. The median number of VTs induced in each patient was 2 (range: 0 to 6) and median cycle length of the clinical VT was 396 ms. A total of 192 VTs were mapped with the mechanism determined to be scar-related re-entry in the majority (90.3%). Less common was inflow cannula-adjacent VT, which was the cause in only 19.3% of VTs post-LVAD implantation, followed by focal or micro re-entry (4.8%) or bundle branch re-entry VT (3.8%). A schematic representation of this is shown in Figure 3.
Acute procedural outcomes
In the 8 included case series (n = 114), the endpoint of noninducible clinical VT was achieved in 77.9% (95% confidence interval [CI]: 68.7% to 4.9%; I2 = 0%); noninducibility of any VT was achieved in 66.6% (95% CI: 42.8% to 83.4%; I2 = 67%). In 10 case reports, the endpoint of noninducible VT was achieved in 9 of 10 (90%). When data were available for analysis, termination of VT storm was achievable in 18 of 20 (90%) cases.
Acute procedural complications
In 8 case series (n = 100), pooled incidence of procedural complications was 9.4% (95% CI: 5.0% to 17.2%; I2 = 0%). In 10 case reports, there were 2 incidences of procedural complications. In detail, minor complications occurred in 4.4% patients, all with groin hematomas, and major complications in 5.5%, with 2 cases of groin pseudoaneurysm requiring surgical repair, 2 cerebrovascular events, and cardiogenic shock in 1 patient. There were 2 cases of LVAD-associated EAM interference and no instances of catheter entrapment in the inflow cannula system. A further description of complications is provided in Table 4.
Mean follow-up for the case series alone (n = 33) was 263.5 ± 267.1 days, and for the 8 case reports the mean follow-up was 188.6 ± 154.1 days.
Any VT recurred in follow-up in 43.6% of patients (95% CI: 28.1% to 60.4%; I2 = 37%) and 3 of 9 (33%) patients in the case reports. The rate of any recurrent VT was 18.8 per 100 person-years (95% CI: 4.0 to 33.5) and 3 of 9 (33%) patients in the case reports. Overall, 10% patients underwent redo catheter ablation, with noninducible clinical VT in 64%. Of these, 2 of the cases had an epicardial origin, 1 of which underwent a redo lower-edge sternotomy and successful epicardial ablation around the LVAD inflow cannula (using an open irrigated radiofrequency catheter in a deflectable sheath with a maximum of 30 W power for 60 to 210 s) the day following the endocardial procedure. Details regarding the mechanisms in those with recurrence and their relationship to the original procedure were not available. In follow-up, there was a significant reduction in ICD therapies or shocks (57.1% vs. 23.8%).
Overall, there were 38 deaths post-LVAD ablation in the follow-up period. The pooled mortality in 7 case series involving 100 patients was 48.1% (95% CI: 26.8% to 70.1%; I2 = 60%); with a mortality rate of 63.5 per 100 person-years (95% CI: 19.1 to 107.9). In the 9 case reports, there was 1 death. A summary of cause of death is provided in Table 5. Importantly, none of the deaths were related directly to VA. The most common causes of death were sepsis, multiorgan failure, thromboembolism, and RV failure.
In 7 case series involving 79 patients, the pooled incidence of cardiac transplantation was 23.4% (95% CI: 14.1% to 36.2%; I2 = 3%; rate of transplantation of 8.7 per 100 person-years; 95% CI: 1.9 to 15.5). In the 9 case reports, 2 patients would go on to have a cardiac transplant (at 35 and 150 days post-VT ablation).
This systematic review summarizes a number of key points in the published LVAD literature:
1. Catheter ablation of VT is feasible as a treatment of “last resort” in patients with LVAD and VT (often with VT storm) refractory to ICD therapies and multiple antiarrhythmic drugs in whom other treatment options are otherwise limited. Ablation allows control of VT with reduction of ICD shocks and termination of VT storm in the majority, at a comparable acute procedural success rate and medium-term recurrence of VT to that of a non–LVAD-related population with VT (5).
2. In contrast to the use of predominantly substrate-guided VT ablation approach in contemporary practice, the hemodynamic tolerability of VT supported by an LVAD allows activation and entrainment to be the predominant mapping strategy.
3. Scar-related re-entry was the predominant arrhythmogenic mechanism in VAD-associated VT; the mechanism is invariably related to scar formed as a result of intrinsic disease with cannula-related VT accounting for ∼20% of all VTs.
4. Of the small number of cases for analysis, the 12-lead ECG of VT could not reliably be used as a predictor of site of putative VT exit.
5. Perceived difficulties with mapping of LVAD-related VT were rare and had minimal or no impact on the procedural efficacy such as smaller left-sided chambers affording limited catheter maneuverability, catheter entrapment within the inflow cannula, and EAM interference.
6. Acute procedural complications were similar to that of the non–LVAD-related population with VT; importantly, none of the reported complications were directly related to the LVAD or VA.
7. Need for cardiac transplantation and mortality were expectedly high in this population, underscoring the advanced disease substrate and high-risk population in whom complex VT ablation procedures were undertaken.
It is critical to note that the published reports came from high-volume, experienced centers for VT ablation, and thus the findings reported in this review may not necessarily be generalized.
Incidence and prognostic implications of VAD-related VT
This review summarizes the published evidence on the natural history, procedural characteristics and outcomes in patients having undergone VAD-related VT ablation. VA is one of the most common complications following LVAD implantation (6). The incidence appears to be declining with the exclusive use of new-generation continuous-flow pumps, but it still remains high, with the most recent estimate of 4.14 arrhythmia events per patient-year of follow-up (7,8). Consistent with this observation is that the vast majority of patients included in this systematic review had continuous flow pumps and only 1 patient with a pulsatile device.
VAs that develop post-VAD substantially increase mortality (2). In a recent retrospective study of 149 patients with second-generation LVADs, VA post-implantation was associated with a 7-fold increase in all-cause mortality (9). Whether mortality is directly caused by onset of arrhythmia or is a marker of global deterioration in cardiac function is unknown (10).
Putative mechanisms of VAD-related VT
Multiple mechanisms have been suggested to explain development of VT in LVAD patients. Electrolyte shifts in the peri-implant period may trigger VA. Alterations in autonomic tone (namely a hypersympathetic state) as a consequence of decompensated heart failure, and concurrent inotropic and vasopressor support can promote triggers for VA. LVAD insertion may alter expression of genes potentially involved in arrhythmogenesis; indeed up-regulation of sarcomeric, calcium handling, and fetal hypertrophic genes have been described, in addition to downregulation of connexin-43, sodium or potassium ATPase, and Kv4.3 channel genes (11–13).
The insertion of a VAD within an advanced cardiomyopathic substrate likely creates unique hemodynamic and electrophysiologic perturbations that contribute to ventricular arrhythmogenesis. New regions of scar may form in response to suture lines related to apical placement of the inflow cannula, dramatic LV unloading and volume reduction, and autonomic and hemodynamic alteration from inotropic support in the peri-implantation period. It is therefore conceivable that new mechanisms or sites are responsible for arrhythmogenesis post-VAD compared with those that occurred before implantation.
Creation of new apical scar promotes creation of a proarrhythmic substrate due to profibrotic remodeling around the inflow cannula insertion site, which may facilitate re-entrant VT (1). Typically, a felt ring is sutured in place of the removed apical wedge, which allows placement of the inflow cannula. In a large series, histological examination of the removed apex revealed interstitial myocardial fibrosis and inflammatory infiltrate in the majority of cases (14), creating ideal substrate for re-entry around an anatomical barrier.
Other mechanisms have also been implicated in the development of VAs post-VAD. In circumstances when the LVAD unloading exceeds the preload capacity (e.g., aggressive over-diuresis or volume depletion from bleeding) a suction effect can occur resulting in the inflow cannula directly contacting the septum or free LV wall inciting mechanical irritation that may trigger VT. Sudden onset monomorphic VT has been shown to correlate to suction events due to excessive LV unloading as determined through VAD monitoring systems (15). Episodes were alleviated by reversal of suction suggesting the events could not be triggered by rapid electrolyte shifts alone. It is also plausible that acute subendocardial ischemia from the surgical insult and repetitive suction events may alter refractory periods and contribute to the proarrhythmic milieu (16).
PVC-triggered VT or ventricular fibrillation is a potential mechanism that could be implicated to VT induction. We observed PVC triggers initiating VT or ventricular fibrillation only rarely in the present systematic review (2.7%), consistent with others (17). Typically, VAs post-VAD are almost exclusively monomorphic VT (as opposed to polymorphic VT or ventricular fibrillation) suggesting re-entry as the predominant underlying mechanism (1,10). In support of our finding that triggered ectopy is uncommon, others have shown VT to proceed regular sinus rhythm or a short-long-short sequence in 89% of episodes. Those that were initiated by a ventricular ectopy were most commonly a ventricular run (46%) followed by PVC (28%) and sudden onset (11%) (10). Additionally, VT cycle lengths post-VAD have been shown to be faster compared with VAs recorded preoperatively (1). Of course, local and systemic changes in the postoperative period may account for these faster rates before assumptions made of different circuits or sites.
In this review, scar-related re-entry was the predominant mechanism in over 90% of VTs (Figure 3). The major implication of this finding is that close catheter proximity to the inflow cannula and ablation around fresh suture lines can be avoided in the majority of cases. Supportive evidence of this mechanism has been through work utilizing a hybrid empirical intraoperative cryoablation to intrinsic scar at the time of LVAD placement to reduce post-procedural VAs (18).
A recent study (19) has also found a disproportionally higher incidence of bundle-branch re-entry, whereas this was the least common mechanism our combined meta-analysis. Advanced cardiomyopathy often with associated conduction disease in the post-VAD heart is the archetypal model for the development of macro-re-entry within the His-Purkinje system. The observation that a typical left bundle branch block VT morphology is common to this group is a useful pre-procedural clue to this being the mechanism. Others have highlighted that when bundle branch re-entry tachycardia occurs in LVAD patients with a right bundle branch block it is atypical with a terminal posterior or leftward axis (S-wave in V1 and R’-wave in V6) due to conduction block from the septum to the free LV wall (20).
VT ECG morphology correlates poorly to VT exit sites
The site of origin of VT using the ECG in structural heart disease is well described (21); however, the same principles may not apply to LVAD-related VT. We reviewed the cases in which the ECG morphology was available to compare the site of successful ablation and regions of scar. Among the 11 cases identified, the clinical VT in all except the BBR case had RBBB morphology. Of interest, 45% had 12-lead ECGs of VT that did not correspond to ablation sites using conventional algorithms. For example, patient 5 had an inferior frontal axis with a focal VT ablated apically at the endocardial-cannula interface. Furthermore, patient 8 had a right bundle branch block morphology that would typically localize to the LV, but was successfully ablated at the RVAD cannula site. This has important implications for pre-procedural planning. Potential explanations for the lack of correlation beyond scar location could be development of progressive diffuse myopathy in the setting of intrinsic scar, anatomical distortion from VAD placement, hemodynamic effects of the VAD, and LV decompression. Further work is required to collaborate this finding.
Major procedural complications are uncommon
The low rate of major complications is in part explained by only experienced operators performing these high-risk cases. The rate of procedural complications is not dissimilar to a non–LVAD-related VT cohort undergoing catheter ablation (6.7% in ischemic cardiomyopathy, 8.3% in nonischemic cardiomyopathy) (5), suggesting that VT ablation in the LVAD population appears to have a reasonable safety profile despite being performed in the highest-risk group of patients.
Practical considerations for VAD-related VT ablation
Type of LVAD
The 2 durable LVADs described in the studies included in this meta-analysis are the HeartMate II and HeartWare. The HeartMate 3 (St. Jude Medical) was recently been compared with the HeartMate II in a randomized control trial and found to have improved outcomes (22). Comparative studies have demonstrated similar hemodynamic profiles (23), incidence, and outcomes of VT (24), despite significant differences in device mechanical design (25). The HeartMate II uses axial flow in which the rotating impeller contains a magnet powered by an electromagnetic motor, rotating on blood-lubricated bearings. Both the HeartWare and HeartMate 3 device use centrifugal flow in which the impeller uses a hybrid suspension mechanism, incorporating passive magnets and hydrodynamic thrust bearings to pump blood out. Compared with pulsatile LVADs, the diameter of the inflow cannula is small and longer in length, giving greater distance to rotor components. Thus, the risk of catheter entrapment is negligible. The HeartWare inflow cannula is implanted in the ventricular apex and is 25 mm in length with a rounded (chamfered) tip, which is 21 mm in diameter. The pump is implanted into the ventricular apex (26). The HeartMate II has a larger diameter of 25 mm, but longer cannula (including a 20-mm inflow conduit) with an acute angulation before entering the motor (Figure 4). The blood contacting surface has a textured microsphere surface, whereas internally it is a polished titanium surface (27). Unlike pulsatile LVADs, continuous-flow LVADs have a consistently phasic (slightly pulsatile) low-velocity inflow and outflow patterns with lower peak velocities (1.5 m/s compared with 2 m/s) (28). The anatomical placement of the inflow cannula is the same for both platforms, aligned with the mitral valve and away from the interventricular septum (29).
Periprocedural planning and monitoring
Catheter ablation of VT in the presence of an LVAD poses many unique challenges predominantly relating to access, mapping, and the potential for catheter entrapment. Ablation requires considerable planning with a team consisting of an advanced heart failure physician (with expertise in LVAD monitoring), electrophysiologists and technicians, anesthetist, and intensivist. A VAD-assisted VT ablation is a long procedure (typically more than 6 h) (30), and these patients are a high-risk group for infection. Clearly defined guidelines for the use prophylactic antibiotics in this group are not available; however, a beta-lactam (e.g., first-generation cephalosporin) is recommended for primary prophylaxis, supplemented by vancomycin where multiresistant Staphylococcus aureus infection is suspected. Intraprocedural monitoring consists of direct central venous pressure tracing and invasive arterial pressure (or left atrial pressure). It must be mentioned that conventional definitions of intravascular and cerebral perfusion pressure required to maintain organ and brain function may not be accurate in patients with nonpulsatile flow (31). Furthermore, these parameters are influenced by the depth of anesthesia and level of inotropic support. Therefore, additional markers of end-organ perfusion are crucial to monitor during the case. A rise in serum lactate, oliguria, increase in pulmonary capillary wedge pressure, right atrial pressure, or sustained hypotension indicate organ underperfusion and should lead to implementation of inotrope or vasopressor therapy. LVAD function relies heavily on RV function. Negative inotropic agents should be avoided, as they can contribute to RV dysfunction by impairing RV output from increasing pulmonary vascular resistance (due to hypoxemia, hypercarbia, and acidosis). An increase in central venous pressure with RV dilatation (reflecting elevated pulmonary arterial pressure) paralleling LVAD output reduction should prompt suspicion for RV failure and consideration of positive inotropic agents (dobutamine and milrinone) or pulmonary vasodilators (e.g., inhaled nitric oxide) (32,33). The main advantage of dobutamine is the less hypotensive and vasodilatory effect than milrinone (34). The use of noninvasive cerebral tissue oxygen saturation has been shown to be a useful adjunct for nonpulsatile mechanical support during VT ablation, adopting a cutoff value to terminate the case if the cerebral tissue oxygen saturation <55% (35). Intraprocedural transesophageal or intracardiac echocardiography is essential, as it not only provides real-time imaging to guide catheter positions and RV assessment, but also monitors for potential complications such as a pericardial effusion. The high volume of catheter irrigation fluid will also increase intravascular volume intraprocedurally, and diuresis with intravenous frusemide will often be necessary.
VAD recipients have a much higher incidence of both bleeding and thrombosis. Acquired coagulopathy (due to the breakdown of large-multimer von Willebrand factor multimers from mechanical shearing and reduction in function activity of the von Willebrand factor) is common (36), paralleled with a prothrombotic state in which patients are particularly at risk of LVAD thrombosis in the early stages post-implantation (37). Current guidelines (38) dictate that all patients with durable continuous-flow devices be anticoagulated with warfarin but the dosing and level may differ depending on the center and device type. Additionally, aspirin and at times a second antiplatelet agent is typically administered concomitantly. The anticoagulation approach in a VAD-assisted VT ablation is dependent on: 1) patients’ pre-existing anticoagulation regime; and 2) the mapping strategy. In the largest case series in this meta-analysis, Sacher et al. (39) performed the ablation with patients fully anticoagulated (international normalized ratio 2 to 3). A pre-procedural interruption of anticoagulation is not recommended due to the thrombotic risk, hence making arterial access (RA) a less favorable approach. In any case, the aortic valve may not open with cardiac contraction, may become fused over time or even be over sewn post-LVAD insertion (40). Considerable care should be taken during the TS access. The LVAD suction effect reduces the left atrial area and adds complexity to TS access (41). It is common practice to temporarily increase preload to facilitate atrial distension and TS access, which may reduce risk related to TS access. LV volumes may be dramatically reduced due to unloading of the LV, potentially making catheter manipulation more difficult, particularly the region around the apical inflow cannula. Contact artefact with impedance drops have also been seen with direct contact of the mapping catheter on the VAD (42,43). The use of intracardiac echocardiography is strongly recommended to assist catheter position and contact (44,45); however, it was only used in 1.8% in the entire cohort.
In those with a subtherapeutic international normalized ratio, systemic therapeutic anticoagulation with intravenous unfractionated heparin targeting an activated clotting time >300 s (depending on the center) should be administered after a transesophageal echocardiogram confirms the absence of a pericardial effusion (for a TS approach) or an uncomplicated ultrasound and fluoroscopic-guided femoral arterial access and sheath positioning (for a RA approach). This step is critical, given the reduced margin of error with the TS puncture with reduced LA size as well as continuous VAD flow and absent femoral arterial pulse pressure. Activated clotting time should be carefully monitored intraprocedurally (every 30 min) and strictly maintained to small fluctuations.
Concern has been raised regarding LVAD-related EAM interference and catheter entrapment. There were no reports of catheter entrapment or damage to the pump in any of the published cases. EAM interference was also uncommon. VAD-related VT cases are highly reliant on EAM. This is due to the complex and often distorted anatomy as well as fluoroscopic barriers (e.g., VAD inflow cannula, drive line, implanted pacing and ICD devices, previous leads, procedural pacing catheters, and a TS sheath) (Figure 5). CARTO (Biosense Webster, Diamond Bar, California) is a magnet-based electroanatomical system that uses a low-level magnetic field created by posteriorly positioned back coils and a location sensor on the catheter tip. The high-speed revolutions of the pump can result in electroanatomical interference (EMI) when a magnet-based EAM is used. A similar ICD-LVAD interaction has been observed in which EMI can be removed by revising the ICD lead away from the inflow cannula to a septal position (46). EAM-EMI has been reported in the setting of both axial and centrifugal LVADs (35,47) including biventricular assist devices (48) and percutaneous devices (49). It can occur during sinus rhythm or paced rhythm (substrate mapping), or during VT or ventricular fibrillation (activation and entrainment mapping). The interference can impede accurate catheter location (loss of catheter visualization), electroanatomical point acquisition, vector orientation, and contact force readings. It typically occurs when the catheter is in close proximity to the apical inflow cannula. This distance is reported to be approximately 4 cm from the VAD (at a pump speed of 6,000 rpm) (50). EMI has not been reported using purely impedance-based systems (Velocity, Abbott, St. Paul, Minnesota) (39), externally placed pumps (VA extracorporeal membrane oxygenation), or TandemHeart (CardiacAssist, Pittsburgh, Pennsylvania) (51). Furthermore, it has not been reported with intracardiac echocardiography integration system. Although EMI rarely prohibits mapping, potential solutions include: 1) ensuring the mapping reference patches are positioned away from the inflow cannula (lower on the patient’s back); 2) using an electrical impedance-only mapping; and 3) reducing the speed of the VAD. Using a percutaneous VAD (Impella microaxial flow device, Abiomed, Danvers, Massachusetts), Vaidya et al. (49) systemically assessed the influence of the 2 systems on EMI. At maximum performance (1.9 to 2.5 l/min), there was no EMI using an impedance-based system (EnSite NavX, St. Jude Medical, St. Paul, Minnesota). However, severe EMI was seen in almost 10% using magnet-based mapping (less if epicardial access). The mean distance from the VAD and severe EMI was 40.1 ± 16.8 mm—irrespective of the mapping chamber. Reducing the flow rate from 1.9 to 2.5 l/min (maximal) to 1.4 to 2.0 l/min (45,000 rpm) eliminated severe EMI in all cases. The “dose-dependent” relationship of motor speed and EMI is reported by others, leading to the conclusion that the electrical frequency applied for rotation on an Impella reduced to 45,000 rpm is filtered out by magnet-based mapping systems (35,47,52). However, the ideal reduced speed (if tolerated) to eliminate EMI in the setting of durable LVADs needs to be further evaluated.
Although successful cases of a modified epicardial approach to VAD-related VT ablation is reported (53), the direct subxiphoid position of the LVAD and extensive fibrotic adhesions precludes a conventional percutaneous approach (54). This is particularly important to consider in patients with probable epicardial substrate, such as those with nonischemic cardiomyopathy being considered for catheter ablation.
High-frequency noise interference can also occur on the both the surface and intracardiac electrogram tracings, potentially impairing ECG morphology discrimination and the accuracy of pace-map matching (35). This limitation is seen across all VAD platforms (HeartMate and Heart Ware), but may occur more frequently with newer-generation devices (55). Here, noise interference is noted to disappear during the “pulse” delivery (every 2 s) from the HeartMate 3 (55). In our own observations, the positions of the ECG cables or amplifier, or deviating the standard electrode positions away from the LVAD do not improve noise. However, significant improvement can be achieved by reducing the low-pass filter of the surface ECG from 100 or 150 Hz to 40 Hz (Figure 6).
All studies were nonrandomized, retrospective, predominantly small case series or single reports, and the results are subject to confounding factors. Reporting bias is likely to have influenced the results. Nevertheless, VT ablation in the high-risk LVAD cohort is complex and only a limited number of patients undergo this procedure at expert centers. Therefore, this review provides the best currently available evidence to highlight the VT characteristics, outcomes, and safety of catheter ablation in this group. It is important to recognize that this analysis is based on a heterogeneous sample in which patients could have distinctly different natural histories and outcomes following ablation. Development of an international registry is needed to provide additional insights into mechanism and treatment of VA in this population.
Catheter ablation is effective in controlling VT post-LVAD insertion in a population in which ablation is often employed as a treatment of last resort when defibrillator and drug therapy have failed. Ablation was particularly effective in acute storm termination and overall resulted in reduction in defibrillator shocks. Acute procedural success, complications, and recurrences are comparable to a non–LVAD-related VT population. Commonly perceived technical barriers such as catheter entrapment are rare and have minimal or no impact on procedural outcome. Scar-related re-entry within pre-existing regions is the predominant mechanism of VT with cannula-related VT responsible for a minority. Rates of transplantation and mortality are high, highlighting the high-risk population in whom ablation is performed.
COMPETENCY IN MEDICAL KNOWLEDGE: The arrhythmogenic mechanisms and characteristics underpinning the development of VT following implantation of VADs are not fully elucidated. Furthermore, procedural outcomes of catheter ablation in this group are not known. This systematic review reveals that scar-related re-entry is the mechanism in the majority of cases. Acute procedural success is high and complication rates are acceptable, but recurrent VT is common. The short-term mortality is significant, but causes of death are unrelated to VA.
TRANSLATIONAL OUTLOOK 1: Catheter ablation for VAD-associated VT is a high-risk group that requires careful planning and is typically limited to experienced centers making reporting of natural history and outcomes uncommon.
TRANSLATIONAL OUTLOOK 2: An international registry is needed to further evaluate the procedural characteristics and clinical outcomes in this group.
Dr. Anderson has been supported by postgraduate scholarships co-funded by the National Health and Medical Research Council (NHMRC) and Royal Australasian College of Physicians NHMRC Woodcock Scholarships. Dr. Hayward has served as a consultant for Medtronic. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
All authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’ institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information, visit the JACC: Clinical Electrophysiology author instructions page.
- Abbreviations and Acronyms
- confidence interval
- electroanatomical interference
- implantable cardioverter-defibrillator
- left ventricular assist device
- premature ventricular complex
- retrograde aortic
- right ventricular assist device
- ventricular arrhythmia
- ventricular assist device
- ventricular tachycardia
- Received June 12, 2018.
- Revision received August 7, 2018.
- Accepted August 17, 2018.
- Ziv O.,
- Dizon J.,
- Thosani A.,
- Naka Y.,
- Magnano A.R.,
- Garan H.
- Kumar S.,
- Romero J.,
- Mehta N.K.,
- et al.
- Greet B.D.,
- Pujara D.,
- Burkland D.,
- et al.
- Yoruk A.,
- Sherazi S.,
- Massey H.T.,
- et al.
- Yaksh A.,
- Kik C.,
- Knops P.,
- et al.
- Chaudhary K.W.,
- Rossman E.I.,
- Piacentino V. 3rd.,
- et al.
- Strecker T.,
- Rosch J.,
- Weyand M.,
- Agaimy A.
- Ma L.,
- Wang L.
- Moss J.D.,
- Flatley E.E.,
- Beaser A.D.,
- et al.
- Noheria A.,
- Mulpuru S.K.,
- Noseworthy P.A.,
- Asirvatham S.J.
- de Riva M.,
- Watanabe M.,
- Zeppenfeld K.
- Mehra M.R.,
- Naka Y.,
- Uriel N.,
- et al.
- Sen A.,
- Larson J.S.,
- Kashani K.B.,
- et al.
- Hosseinipour M.,
- Gupta R.,
- Bonnell M.,
- Elahinia M.
- Estep J.D.,
- Stainback R.F.,
- Little S.H.,
- Torre G.,
- Zoghbi W.A.
- Uriel N.,
- Pak S.W.,
- Jorde U.P.,
- et al.
- Cook J.L.,
- Colvin M.,
- Francis G.S.,
- et al.
- Sacher F.,
- Reichlin T.,
- Zado E.S.,
- et al.
- Schechter M.A.,
- Joseph J.T.,
- Krishnamoorthy A.,
- et al.
- Nakahara S.,
- Chien C.,
- Gelow J.,
- et al.
- Pedretti S.,
- Cipriani M.,
- Bonacina E.,
- et al.
- Mikovcak J.,
- Neuwirth R.,
- Stritecky J.,
- Branny M.
- Thosani A.J.,
- Bailey S.H.,
- Raina A.
- Miller M.A.,
- Dukkipati S.R.,
- Chinitz J.S.,
- et al.
- Ho G.T.H.,
- Urey M.A.,
- Adler E.D.,
- Pretorius V.G.,
- Hsu J.C.
- Vaidya V.R.,
- Desimone C.V.,
- Madhavan M.,
- et al.
- Higgins S.L.H.K.,
- Meyer D.,
- Pless T.
- Miller A.C.,
- Evans A.C.,
- Revenaugh J.,
- Weiss J.P.,
- Reid B.B.,
- Kfoury A.G.
- Miller M.A.,
- Dukkipati S.R.,
- Mittnacht A.J.,
- et al.
- Whang W.,
- Patel M.R.,
- Iyer V.,
- et al.
- Hottigoudar R.U.,
- Deam A.G.,
- Slaughter M.S.,
- et al.
- Batul S.A.,
- Gopinathannair R.
- Snipelisky D.,
- Reddy Y.N.,
- Manocha K.,
- et al.
- Szegedi N.,
- Szeplaki G.,
- Merkely B.,
- Geller L.
- Romano G.,
- Raffa G.M.,
- Ruggieri A.,
- et al.
- Schade A.,
- Nentwich K.,
- Deneke T.
- Mulukutla V.,
- Lam W.,
- Simpson L.,
- Mathuria N.
- Ellis C.