Deep sedation for transvenous lead extraction: a large single-centre experience

Abstract

Aims

Transvenous lead extraction for cardiac implantable electronic devices (CIED) is of growing importance. Nevertheless, the optimal anaesthetic approach, general anaesthesia vs. deep sedation (DS), remains unresolved. We describe our tertiary centre experience of the feasibility and safety of DS.

Methods and results

Extraction procedures were performed in the electrophysiology (EP) laboratory by two experienced electrophysiologists. We used intravenous Fentanyl, Midazolam, and Propofol for DS. A stepwise approach with locking stylets, dilator sheaths, and mechanical sheaths via subclavian, femoral, or internal jugular venous access was utilized. Patient characteristics and procedural data were collected. Logistic regression models were used to identify parameters associated with sedation-related complications. Extraction of 476 leads (dwelling time/patient 88 ± 49 months, 30% ICD leads) was performed in 220 patients (64 ± 17 years, 80% male). Deep sedation was initiated with bolus administration of Fentanyl, Midazolam, and Propofol; mean doses 0.34 ± 0.12 μg/kg, 24.3 ± 6.8 μg/kg, and 0.26 ± 0.13 mg/kg, respectively. Deep sedation was maintained with continuous Propofol infusion (initial dose 3.7 ± 1.1 mg/kg/h; subsequently increased to 4.7 ± 1.2 mg/kg/h with 3.9 ± 2.6 adjustments) and boluses of Midazolam and Fentanyl as indicated. Sedation-related episodes of hypotension, requiring vasopressors, and hypoxia, requiring additional airway management, occurred in 25 (11.4%) and 5 (2.3%) patients, respectively. These were managed without adverse consequences. Five patients (2.3%) experienced major intraprocedural complications; there were no procedure-related deaths. All of our logistic regression models indicated intraprocedural support was associated with administration higher Fentanyl doses.

Conclusion

Transvenous lead extraction under DS in the EP laboratory is a safe procedure with high success rates when performed by experienced staff.

Introduction

Complications associated with cardiac implantable electronic devices (CIED) have increased with recent increases in the number of CIED implantations.¹ CIED infection and lead malfunction are the main reasons for transvenous lead extractions (TLEs). The patient population that typically requires such extraction is often high risk; older patients, those with multiple comorbidities, and those with leads implanted many years ago. According to the 2017 Heart Rhythm Society consensus statement on lead management,² many centres perform lead extractions under general anaesthesia (GA) to minimize patient discomfort and facilitate use of intraprocedural transesophageal echocardiography (TEE). GA also eliminates the need for emergency intubation should complications occur and consequently allows the anaesthesia team to focus on resuscitation rather than intubation.² However, the European lead extraction study (ELECTRa)—a registry of the European Heart Rhythm Association (EHRA) affiliated centres—reported relatively uniform distribution of anaesthesia approaches (performed mainly by non-cardiothoracic surgeons); 39% general, 31% local, and 31% sedation.³ Individual studies reporting the outcome of lead extraction sometimes describe the anaesthesia approach and these typically reflect the ELECTRA-experience. Clearly, procedures performed in cardiac electrophysiology (EP) laboratories are more likely to employ non-general anaesthesia approaches,^4–7 whereas in operating rooms (ORs), GA is typically used.^4,8,9

Safe administration of deep sedation (DS) by interventionalists, or dedicated nurses, has been evaluated for ablation of atrial fibrillation (AF),^10–12 implantation of cardioverter-defibrillators (ICDs),¹³ and also in other medical fields; e.g. in gastro-intestinal endoscopic procedures.¹⁴ A few head-to-head comparisons between GA and DS in cardiac procedures indicated similar safety profiles and potentially simplified procedures for the latter.^4,15,16

The duration of CIED extraction procedures together with the associated pain and requirement for immobility means that intraprocedural sedation is widely accepted. Nevertheless, there is ongoing debate whether GA or DS is preferable. The lack of data on the use of DS in TLE procedures prompted us to assess the feasibility and safety of this approach in a single-centre cohort from a tertiary referral hospital.

Methods

Patient population

Over 2 years, 250 consecutive patients were referred to our centre for TLE because of infection or lead malfunction. All patients underwent complete cardiac evaluation, which included a comprehensive medical history and assessment of perioperative risk-classification according to the American Society of Anesthesiologists (ASA, physical status classification system¹⁷) In addition, we conducted laboratory tests, 12-lead electrocardiogram (ECG), CIED interrogation, transthoracic echocardiography (TTE), and, in cases of infection, TEE. We obtained written informed-consent for all procedures and the data evaluation was approved by our institution’s ethics committee and in accordance with the declaration of Helsinki.

Patient monitoring and sedation

All procedures were performed in the EP laboratory by two experienced electrophysiologists. Rapid-response back-up from the anaesthesiology and cardiothoracic surgery departments was always available. Both electrophysiologists were trained in advanced life support, airway management, application of anaesthetic drugs, and intensive care medicine. Two nurses assisted; one monitored the patient’s vital signs and administered intravenous (iv) drugs under direct supervision of the operating electrophysiologist (proceduralist-directed nurse-administered model). All the nurses had basic life-support skills (many possessed advanced training) and had worked in intensive care or as anaesthesia nurses. Patients were monitored with continuous ECG, invasive blood pressure, oxygen saturation, and frequent arterial blood gas analyses. All patients received two peripheral vein cannulas and one or two femoral venous sheaths in preparation for the procedure. The level of sedation was classified according to the ASA¹⁸: DS was defined as being unresponsive to vocal stimuli, tolerating an oropharyngeal airway, and breathing spontaneously. Sedation was initiated with iv boluses of Fentanyl, Midazolam, and Propofol and maintained with continuous iv administration of Propofol and additional boluses of Midazolam and/or Fentanyl as indicated. Fentanyl was used for analgesia and not to aid lead extraction by venodilation. Doses were adjusted as indicated by the patients’ haemodynamic and sedation status. If necessary, patients received saline infusion, Cafedrine/Theodrenalin boluses, or continuous Norepinephrine to achieve and maintain systolic pressure above 90 mmHg. Oxygen was applied via an oxygen cannula and flow adjusted to achieve a saturation level >90%. If peripheral oxygen saturation decreased below 90% and was unresponsive to increased oxygen flow or repositioning of the head and neck, patients were ventilated by face mask or laryngeal mask. Endotracheal intubation, if needed, was performed by an anaesthesiologist.

Lead extraction

Local anaesthesia with 1% Mepivacaine was administered at the intended surgical location. Pacemaker-dependent patients received a temporary right ventricular pacing catheter via one of the femoral sheaths. After opening the device pocket, leads were exposed up to their sleeves and untied. If manual traction was unsuccessful, a systematic approach with ongoing risk-benefit analysis was applied using locking stylets (Liberator^®, Cook Medical, Bloomington, IN, USA), sutures, and mechanical dilation polypropylene sheaths (Byrd Dilator Sheaths, Cook Medical, Bloomington, IN, USA). When necessary, a mechanically controlled rotation sheath (Evolution^®, Cook Medical, Bloomington, IN, USA), snares, or both were used via subclavian, femoral, or internal jugular venous access.

Pacemaker-dependent patients received a transcutaneous screw-in pacemaker lead connected to an external pacemaker after removal of the hardware if re-implantation had to be postponed for medical reasons. Procedure duration was defined as the time between first incision and the last skin suture.

We defined endpoints based on intention-to-treat analysis. Complications and success rates were defined according to Heart Rhythm Society and EHRA guidelines for TLEs.^2,19

Statistical analysis

Continuous variables were presented as mean with standard deviation and 95% confidence interval, or median with interquartile range, and categorical variables were presented as frequency and percentage. Comparisons of patient and procedural characteristics were performed using the Student’s t-test for continuous and χ² test for categorical variables.

We sought to identify parameters associated with DS-related need for support. Deep sedation-related hypotension and hypoxia was assumed when other reasons for hypotension or hypoxia (i.e. pericardial tamponade, bleeding, pneumothorax, traction on the leads) were excluded. Potential factors were first identified from univariate analysis and then included in logistic regression models to evaluate the likelihood of association. The processes involved in model construction are described in the results.

Statistical analysis was conducted using Excel and Stata 15.1 (StataCorp, College Station, TX, USA).

Results

Two hundred and twenty patients (mean age 64 ± 17 years, 80% male) out of 250 consecutive patients underwent TLE under DS. The 30 patients excluded comprised 21 in whom TLE was performed under conscious sedation, seven who were transferred from the intensive care unit and were already sedated, intubated, and ventilated, one with lead malfunction who was treated in the OR and therefore received GA, and one patient with incomplete records.

Baseline patient characteristics are listed in Table 1. Pacing indications and primary prevention of sudden cardiac death in patients with non-ischaemic or ischaemic cardiomyopathy were the leading causes for initial CIED implantation. Lead malfunction and infections were the most common reasons for lead extractions (Table 1).

Sedation and analgesia-related medication

Procedural characteristics are summarized in Table 2. Deep sedation was initiated with an average bolus injection of Fentanyl (0.34 ± 0.12 μg/kg), Midazolam (24.3 ± 6.8 μg/kg), and Propofol (0.26 ± 0.13 mg/kg). Deep sedation was maintained with an average initial Propofol dose of 3.7 ± 1.1 mg/kg/h; subsequently titrated up to a mean of 4.7 ± 1.2 mg/kg/h with 3.9 ± 2.6 adjustments during the procedure. Additional boluses of Fentanyl and Midazolam were also administered; Fentanyl—average of 1.7 ± 1.4 (median, interquartile range: 1, 1–3) additional boluses and a total cumulative dose of 0.90 ± 0.55 µg/kg; Midazolam—average of 0.7 ± 1.2 (median, interquartile range: 0, 0–1) additional boluses and a total cumulative dose of 36.1 ± 23.3 μg/kg.

Success rate and complications

Of the 476 leads identified for removal, 95.6% (455) were completely extracted. Lead removal with clinical success was achieved for 97.7% (465). Complete procedural success rate and procedural clinical success rate were 91.4% (201/220) and 95.9% (211/220), respectively; a fragment (<4 cm) remained for 10 leads (2.1%) in nine patients. Extraction failed for 11 leads (2.3%) in ten patients. In four of 10 patients with failed extraction, five leads were surgically removed. In the other six, the indication for extraction was lead malfunction not infection. Therefore, the risk-benefit analysis was in favour of leaving the leads in place. There was no association between DS-related events and procedural clinical success.

Four patients developed pericardial tamponade during TLE; diagnosed by decreased blood pressure and decreased excursion of the cardiac silhouette in left anterior oblique fluoroscopic view. Two patients were managed successfully with pericardiocentesis. This was done under DS without need for additional airway support; however, a vasopressor, Cafedrin/Theodrenaline, was given. The other two patients were transferred to the OR, after pericardiocentesis restored hemodynamic stability, but failed to control bleeding. One of these patients was intubated for safety reasons at the discretion of the interventionalist after tamponade occurred in the EP lab. In the OR, tears in the right atrium and the right ventricular free wall, respectively, were repaired using suture. A patient with ischaemic cardiomyopathy (ejection fraction 16%) developed electromechanical dissociation during the attempt to extract a single-coil ICD lead (12 years old) after his two pacemaker leads (12 and 6 years old) had been removed. After a brief period of chest compression, and under Dobutamin application as well as continuous positive airway pressure (CPAP) mask ventilation, the procedure was completed. However, the ICD lead was abandoned. Re-Implantation of a cardiac resynchronization therapy defibrillator was performed 2 days later on the contralateral side and the patient discharged without sequelae. In the above-mentioned five patients, we do not know if they would have required sedation-related support and so they were excluded from comparative analysis. They were, however, included in the ‘all’ column in Tables 1 and 2.

No procedure-related deaths occurred. Three patients died within 1 month of the procedure; at 9, 18, and 30 days, respectively. One of these also experienced intraprocedure tamponade requiring intubation (as described above), and so was excluded from analysis; the other two were included. Two additional patients developed pneumothorax that required drainage the day after the procedure. Because neither patient experienced hypoxia during the procedure they were included in the analysis.

Sedation-related side effects

Episodes of hypotension, attributed to DS that required vasopressors occurred in 11.4% (25) of patients. Twenty-four patients (10.9%) experienced only brief periods of hypotension and were treated with Cafedrin/Theodrenaline (mean dose 0.69 ± 0.68 mg/kg, median dose 0.48 mg/kg) together with reduction of their Propofol infusion. However, one patient required continuous Norepinephrine infusion to maintain a systolic pressure above 90 mmHg.

Oxygen flow averaged 2.4 ± 1.3 L/min and the minimum measured oxygen saturation was 94 ± 3% (range 83–100%). Most patients (96.8%; 213) maintained spontaneous respiration throughout the procedure. However, five patients (2.3%) required intermittent airway management because of sedation-related hypoxia. Two were treated with mask ventilation, two with non-invasive CPAP ventilation by mask, and one patient received a laryngeal mask.

All patients with sedation-related hypotension or hypoxia were managed successfully by the EP team and discharged without sequelae.

Comparison between patients requiring vasopressors due to sedation-related hypotension or airway support due to hypoxia and those without these side-effects revealed several differences in baseline and procedural characteristics (Tables 1 and 2). We examined these in more detail in an attempt to identify parameters associated with the need for support. Univariate analysis identified five potential candidates for inclusion in an initial logistic regression model; total Fentanyl dose (divided into quartiles), the presence of coronary artery disease (CAD), procedure time, lead dwelling time, and the number of leads per patient (the latter two were expressed as categorical parameters—see Table 2). However, there was a correlation between procedure duration and dwelling time (coefficient 0.25, R² = 0.06, P < 0.0001) for the entire cohort, which appeared stronger in patients who required intraprocedural support (coefficient 0.41, R² = 0.20, P = 0.014). Therefore, inclusion of both in a model would introduce collinearity. All four remaining parameters were included in the model, which was adjusted for age (Model 1; Table 3). Because this initial model sought to identify potential factors associated with the need for support, we tested four hypotheses and so a Bonferroni correction was applied. In this case, we divided the original α value of 0.05 by four; the corrected α was 0.0125. Only total fentanyl dose showed weak evidence of association (P = 0.014) and so only this parameter was examined further. Thirty patients required support and therefore subsequent models included only three parameters; to include more would risk over-fitting the models. In Model 2 (Table 3), we tested the hypothesis that total Fentanyl dose was associated with need for support; this model assumed a linear increase in odds ratio across quartiles. The crude model was then adjusted for CAD (the only parameter that produced a change in estimate of ∼10%) and then also for age. This model indicated strong evidence for an association between total Fentanyl dose and the need for support. Model 3 (Table 3) examined the assumption of linearity of odds ratio increase over the dose quartiles. The test for departure from linear trend revealed no departure; although the confidence intervals are wide. All the models indicated an increase in total Fentanyl dose was associated with increased need for support.

Discussion

We demonstrated that DS is a feasible and safe anaesthesia approach in this cohort of 220 patients undergoing TLE. As far as we are aware, this study is the first to show that Propofol sedation with Fentanyl analgesia can be safely administered and monitored by interventionalists without assisted ventilation or an anaesthesiologist present. Even though sedation-related side effects were relatively common (13.6%), they were easily managed. Currently, GA is the recommended approach for TLE procedures; to facilitate intraprocedural TEE and resuscitation if needed.² Nevertheless, there are no comparative outcome data available that justify routine intraprocedural use of TEE in TLE procedures. Although TEE is considered safe, it may result in complications.²⁰ Instantaneously available fluoroscopy together with haemodynamic monitoring can be used as an alternative tool to diagnose pericardial tamponade.²¹ We also used instantaneously available sterile-packed hand-held TTE to confirm tamponade. Nevertheless, some data indicate routine TEE can be used in DS without disadvantage; for example, in patients undergoing percutaneous mitral valve repair (PMVR).¹⁵ Complication rates and mortality associated with lead extraction procedures are low, irrespective of whether the procedures were done in the OR with GA or in the EP lab with DS; for example, as reported in a non-randomized study.⁴

Four patients in our study suffered pericardial tamponade; immediate puncture was the only required intervention in two cases. In the other two patients, pericardial puncture led to haemodynamic stabilization; however, bleeding control was unsuccessful. Nevertheless, uncomplicated transfer to the OR was possible and the necessary surgical treatment was performed. The most serious complication, vascular tears, is associated with 50% mortality even with immediate surgical treatment. There would be insufficient time to transfer such cases to the OR. Nonetheless, deployment of an endovascular balloon (Bridge; Spectranetics Corporation, Colorado Springs, CO, USA) in such cases would reduce mortality considerably.²² We adapted our TLE policy after availability of the Bridge Balloon. Accordingly, all patients with a high risk of complications are now prepared for balloon deployment.²³

In contrast to our promising results, Kutarski et al. observed lower cardiac tamponade-related mortality for patients treated in the OR vs. the EP lab. On the other hand, recent evaluation of the ELECTRA registry indicated a strategy of pericardiocentesis, followed by a rescue-surgical approach appeared effective and safe for tamponade treatment. This supports our findings and highlights the need for appropriate risk stratification before intervention. In summary, the success and complication rates in our cohort are comparable to numerous prior studies.^3,4,9,24

In our cohort, 11.4% (25) of patients required vasoconstrictive medications due to sedation-related hypotension and 2.3% (5) of patients required mask ventilation due to sedation-related hypoxia. GA facilitates airway management; but, is also associated with risk of hypotension. Such complications have occurred in studies of ventricular tachycardia ablation and PMVR too.^15,25 In patients undergoing PMVR, doses of Propofol and Norepinephrine were lower for DS vs. GA.¹⁵ However, high catecholamine doses can be associated with catecholamine-induced cardiotoxicity.²⁶ Therefore, the mode of anaesthesia must be carefully chosen in TLE patients because of the prevalence of severe cardiovascular comorbidities. Our results are consistent with a similar study of DS at our institution in patients undergoing AF ablation; a procedure typically performed in the EP lab with DS, but with similar risk of severe complications such as pericardial tamponade. Short episodes of sedation-related hypotension (requiring vasopressors) and hypoxia (requiring mask ventilation) occurred in 15% and 1% of the AF ablation patients. None of these patients required intubation or the assistance of an anaesthesiologist.¹¹ However, the cohorts were dissimilar; our patients had a higher incidence of heart failure, ischaemic heart disease, diabetes mellitus, lower ejection fraction, shorter procedure times, and slightly lower doses of Midazolam, Fentanyl, and Propofol for anaesthesia initiation and maintenance. Similar experiences with DS for AF ablation were reported from two other German tertiary centres^12,27 and are consistent with the practicality and safety of this approach. Furthermore, patients reported high levels of satisfaction with DS during catheter ablation.²⁸

In a cohort of patients with more comorbidities (85% ASA Classes III and IV) undergoing CIED surgery (cardioverter defibrillator or pacemaker implantation, battery replacement, and lead extraction), the reported incidence of compromising hypoxia and hypotension under DS by anaesthetists were higher; 16% and 15%, respectively.²⁹ Patients in a study of PMVR were also comparable to our patient cohort in terms of comorbidities. A total of 271 consecutive patients underwent PMVR in a comparative study; 72 procedures performed under GA and 199 under DS. The authors observed that, despite the above-mentioned lower doses of Propofol (DS = 369 ± 230 mg vs. GA = 743 ± 228 mg; P < 0.001) and Norepinephrine (DS = 0.2 ± 0.3 mg vs. GA = 1.1 ± 1.6 mg; P < 0.001), procedure and fluoroscopy times, and dose-area product were significantly lower with DS. There was no difference between GA and DS complication rates.¹⁵ Deep sedation was associated with shorter post-intervention stay in the intensive care unit and hospital. No difference in mortality was observed.^15,16,30 Nonetheless, these studies must be interpreted with caution because selection bias may be present. Confirmation of these results awaits assessment in randomized trials.

Patients requiring DS-related support showed a trend towards a higher probability of CAD, extraction of more leads with longer dwelling times and received higher total doses of fentanyl. Patients who received the highest total Fentanyl doses were three and half times more likely to require intraprocedural support vs. those in the lowest dose quartile. Additional study is required to determine the precise magnitude of this effect (the 95% confidence intervals were wide) and if the risk is also increased at lower doses.

What are the potential clinical implications of this apparent association? In our study, all the adverse events were easily managed. Nonetheless, administration of higher Fentanyl doses should prompt increased awareness and caution because these patients will be at higher risk for sedation-related complications, which could lead to devastating consequences if not managed properly in a timely manner.

In our opinion, a well-trained staff can manage the level of sedation to prevent critical persistent hypotension and hypoxia, to avoid transition to GA, and to be able to resolve critical situations if they appear. Propofol is a short-acting anaesthetic with broad use for induction of anaesthesia and sedative agent in intensive care settings. Its rapid onset and short half-life (2–4 min) combined with improved patient satisfaction vs. sedation with narcotics and benzodiazepines and the association with a quick and ‘smooth’ recovery^31,32 enhances its appeal. However, Propafol’s potential to cause rapid changes in neuropsychological function, from conscious sedation to DS, or even to narcosis with cardiovascular and respiratory depression and apnoea, should be considered. The risks of these effects can be mitigated by appropriate dose adjustment and patient management. In the current study, we used an individualized approach with several adjustments of the medications to maintain hemodynamic stability in DS. In our cohort, we found no long-term sedation-associated sequelae.

Limitations

This was a single-centre experience and we did not directly compare the different anaesthesia approaches. Therefore, our results may not be generalizable and additional studies are needed. Our logistic regression models suggest an association between higher total Fentanyl doses and requirement for intraprocedural support. However, the small sample size and consequent limited number of parameters that could reasonably be included in the model indicates the results should be interpreted with caution and be considered ‘hypothesis generating’. Moreover, the wide 95% confidence intervals indicate that although there is an increase in the odds of required support, the magnitude of this increase was poorly defined.

Conclusion

Deep sedation appears safe and feasible for TLE in centres with experienced staff. However, patients’ characteristics, risk predictors, and preferred extraction approaches need to be considered. Additional studies are needed to identify patient groups that might benefit from one anaesthesia mode vs. another.

Conflict of interest: K.B., J.L., A.M., S.R., M.D., G.H., and N.D. report research grants from Abbott, Biotronik, Boston Scientific, and Medtronic to the institution without personal financial benefits. P.W. has nothing to declare.

References

Author notes