The MB score: a new risk stratification index to predict the need for advanced tools in lead extraction procedures

Abstract

Aims

A validated risk stratification schema for transvenous lead extraction (TLE) could improve the management of these procedures. We aimed to derive and validate a scoring system to efficiently predict the need for advanced tools to achieve TLE success.

Methods and results

Between November 2013 and March 2018, 1960 leads were extracted in 973 consecutive TLE procedures in two national referral sites using a stepwise approach. A procedure was defined as advanced extraction if required the use of powered sheaths and/or snares. The study population was a posteriori 1:1 randomized in derivation and validation cohorts. In the derivation cohort, presence of more than two targeted leads (odds ratio [OR] 1.76, P = 0.049), 3-year-old (OR 3.04, P = 0.001), 5-year-old (OR 3.48, P < 0.001), 10-year-old (OR 3.58, P = 0.008) oldest lead, implantable cardioverter-defibrillator (OR 3.84, P < 0.001), and passive fixation lead (OR 1.91, P = 0.032) were selected by a stepwise procedure and constituted the MB score showing a C-statistics of 0.82. In the validation group, the MB score was significantly associated with the risk of advanced extraction (OR 2.40, 95% confidence interval 2.02-2.86, P < 0.001) and showed an increase in event rate with increasing score. A low value (threshold = 1) ensured 100% sensibility and 100% negative predictive value, while a high value (threshold = 5) allowed a specificity of 92.8% and a positive predictive value of 91.9%.

Conclusion

In this study, we developed and tested a simple point-based scoring system able to efficiently identify patients at low and high risk of needing advanced tools during TLE procedures.

Introduction

Lead management of cardiovascular implantable electronic device (CIED) is crucial to ensure optimal patient care and transvenous lead extraction (TLE) represents a growing aspects in this field.¹ Infection, lead dysfunction, lead-related complications, and venous access issues are the main indications for a lead removal procedure.²

Despite TLE showed high success rates and low complication incidences in several studies,^3,4 there is still lack of consensus regarding the appropriate training needed for sites which could safety perform these procedures.⁵

The reason is that TLE includes a wide and increasing spectrum of tools and techniques and it becomes more and more difficult to acquire working knowledge with all of them in performing a limited number of cases.²

In this scenario, a risk stratification schema able to predict the need for advanced techniques to achieve a successful TLE may have an important clinical application.^6,7 On the one hand, it could avoid futile procedures in low-volume centres for complex patients who need dedicated team with complete competency in all available approaches, including minimally invasive surgery or thoracoscopy.⁸ On the other hand, it could identify which patients can be safely extracted in sites with limited experience leading to better resources management. The major obstacle associated to a risk scoring system is that its development and validation requires two independent cohorts, requiring large databases from multicentre registries.

The aim of this multicentre registry was to derive and validate in two independent cohorts of TLE procedures a scoring system to efficiently predict the need for advanced tools to achieve complete procedural success.

Methods

We prospectively collected data from consecutive CIED recipients who underwent TLE procedures in two national referral sites for this procedure between November 2013 and March 2018. Patients and procedural characteristics were obtained prospectively on electronic case form reports. Patients gave informed consent and the registry was approved by the competent ethics committees.

Aim of the analysis

In a real-world cohort of patients, we aimed to develop a scoring system to efficiently predict the need for advanced tools to achieve complete procedural success in a stepwise approach. The risk stratification schema was designed to be based on risk factors which could be easily obtained by leads and implant characteristics. The developed score, defined MB score from the initials of the principal investigators of the two study sites, was then tested in an independent cohort of TLE procedures. Here, we report the results of the derivation analysis and the performances of the new score in the validation analysis.

Definitions

All terms used in the present analysis were defined in accordance with the 2018 expert consensus statement on lead extraction from the European Heart Rhythm Association (EHRA).² Indications for TLE were classified in infection, lead dysfunction, abandoned functional leads, and other indications. The latter class could include venous access issues, access to magnetic resonance imaging, chronic pain, and a number of other rare indications for extraction.

Complete procedural success was defined as removal of all targeted leads and material. All intraprocedural and early post-procedure complications (i.e. those evident within 30 days) were also documented and traced.

The endpoint of our analysis was the use of advanced tools to achieve complete procedural success. This approach represents an objective method to identify the most difficult procedures when a stepwise strategy is applied, as well as allowing considerations on costs.

The following definitions were used:

Simple extraction: complete procedural success achieved with simple traction (manual and/or using standard stylets), locking stylets, or telescoping dilator sheaths.

Advanced extraction: use of powered sheaths (mechanical and/or laser) or snares by femoral route.

Stepwise approach for transvenous lead extraction procedures

All the procedures were performed in the electrophysiological laboratory with cardiac surgery available on-site. In pacing-dependent patients, a temporary pacemaker was placed by femoral route. In case of infection indication, a transoesophageal echocardiogram was routinely performed to assess the presence of vegetations. During the procedure, the operator used a standard stepwise approach which included a transition from simple to more complex strategies (Figure 1). Once the device pocket was opened, the leads were disconnected and prepared for lead extraction by removing the anchors, cutting the proximal portion, and retracting the active fixation mechanism if present. At the beginning of each case, two attempts of gentle manual traction were subsequently performed: the first after the introduction of a regular stylet and the second using a locking stylet. When minor traction revealed that there was more than trivial fibrosis, but the operator considered the fibrous adhesions not so consistent to warrant a direct use of powered tools, a telescoping dilator sheath could be advanced over the lead to try to detach fibrous adhesions.

The extraction was defined simple if the removal of all targeted leads was achieved applying only these first steps. Conversely, if they were ineffective, more advanced techniques were performed, which basically included the use of powered sheaths or, for free-floating portion of leads, snares. Either mechanical and laser sheaths were available in our sites and used at the discretion of the operating physician.

Statistical analysis

Before data analysis, study population was a posteriori 1:1 randomized to derivation and validation cohorts. The randomization was stratified by study endpoint to obtain the same distribution of advanced extractions in the two groups.

In the derivation cohort, the association of several patients and leads characteristics with the study endpoint was estimated by univariate logistic regression models first. Results of logistic analyses were reported in terms of odds ratio (OR) with 95% confidence interval (CI). A multivariate logistic regression model was then applied selecting significant variables by an automatic backward stepwise procedure (P to remove > 0.10). Survival variables were used to create the MB score. The receiver operating characteristic curve was then plotted, reporting the area under the curve (AUC).

The performances of the MB score to predict an advanced extraction were then tested in the independent validation cohort. Sensitivity, specificity, positive and negative predictive values (PPV and NPV) were calculated for several thresholds of the score. The advanced extraction rates according to the derived risk stratification schema were also reported.

Continuous variables were reported with median and interquartile range (IQR), binary or categorical variable as counts and percentages of non-missing data. Comparisons between derivation and validation cohorts were performed with the Mann–Whitney U test for continuous variables, Pearson’s χ² or Fisher’s exact test, as appropriate, for binary variables. Normality was checked with the Shapiro–Wilks test. Statistical significance was defined as P < 0.005. All statistical analyses were performed using STATA 11SE software version (StataCorp LP, TX, USA).

Results

Study population

A total of 973 consecutive CIED recipients who underwent TLE were included in the study. Patients were 70 (IQR 60–77) years old and 77.4% males. Ischaemic (39.8%) and dilated (30.6%) cardiomyopathies were the most prevalent underlying heart diseases. Cardiovascular implantable electronic devices were uniformly distributed among pacemaker (33.1%), implantable cardioverter-defibrillator (ICD) (34.2%), and cardiac resynchronization therapy (32.7%) devices. Infection was the most frequent indication for TLE amounting for 55.8% (with evidence of vegetations in 28.8% of cases), while 35.9% of procedures were due to lead dysfunction. The median number of targeted leads per patient was 2 (IQR 1–3). Specifically, 67.9% and 33.3% of patients had more than two and three leads to be removed, respectively. As shown in Table 1, no significant differences were observed between study cohorts.

Extracted leads, tools and outcomes

Complete procedural success was achieved for 947 (97.3%) patients, resulting in 1960 extracted leads (67.0% pacemaker, 18.6% single-coil and 14.4% dual-coil ICD leads). Of them, 1266 (68.4%) were passively and 585 (31.6%) were actively fixed. The majority of the leads were located in the right ventricle (52.6%), followed by right atrium (32.0%) and coronary venous system (15.4%). Leads belonging to all major manufacturers were extracted: 913 (46.6%) Medtronic (Minneapolis, MN, USA) leads, 469 (23.9%) Abbott (formerly St Jude Medical, Sylmar, CA, USA) leads, 250 (12.7%) Boston Scientific (St Paul, MN, USA) leads, 212 (10.8%) Biotronik (Berlin, Germany) leads, 113 (5.8%) Microport (formerly LivaNova and Sorin, Clamart, France) leads, and 3 Osypka Medical (Berlin, Germany) leads. The median implant time was 5.8 (2.8–8.7) years.

Three hundred and thirty-three TLEs (34.2%) were defined as simple extraction. Among them, telescoping non-powered dilator sheaths were used only in 14 (1.5%) procedures. On the other hand, 640 (65.8%) were advanced extractions as they required powered sheaths and/or snares.

There were 6 (0.6%) major complications, 2 (0.2%) procedure-related deaths, and 4 (0.4%) vascular lacerations requiring surgical revision. Minor complications included transient hypotension that responded to fluid infusion or minor pharmacological intervention in 28 (2.9%) patients, pericardial effusion without medical intervention in 12 (1.2%) patients and sustained ventricular tachycardia requiring electrical cardioversion in 5 (0.5%) patients.

The flowchart of the TLE procedures is shown in Figure 2.

Results of the derivation analysis

In the derivation cohort, several variables were significantly associated with the risk of advanced extraction at the univariate analysis (Table 2). All of them were then included in a multivariate model. The automatic stepwise procedure selected six variables: presence of more than two targeted leads (OR 1.76, 95% CI 1.01–3.11, P = 0.049), 3-year-old (OR 3.04, 95% CI 1.56–5.95, P = 0.001), 5-year-old (OR 3.48, 95% CI 1.83–6.62, P < 0.001), 10-year-old (OR 3.58, 95% CI 1.39–9.23, P = 0.008) oldest lead, ICD lead (OR 3.84, 95% CI 2.30–6.42, P < 0.001), and lead with passive fixation (OR 1.91, 95% CI 1.06–3.46, P = 0.032). Using a point-based scoring system (Figure 3), these variables constituted the MB score, whose unitary increase was associated with an OR of 2.74 (95% CI 2.26–3.31, P < 0.001) for advanced extraction. The C-statistics showed an AUC of 0.82 (Figure 4A).

Results of the validation tests

In the validation group, the procedures across the derived MB score classes were: 12 (2.5%) with Score 0, 42 (8.6%) with Score 1, 75 (15.4%) with Score 2, 99 (20.4%) with Score 3, 110 (22.5%) with Score 4, 121 (24.7%) with Score 5, and 26 (5.3%) with Score 6. The MB score confirmed to be a strong predictor of advanced extraction with an OR of 2.40 (95% CI 2.02–2.86, P < 0.001) and an AUC of 0.80 (Figure 4B). Table 3 summarizes the performances of the MB score as advanced extraction predictor. Different values of sensitivity, specificity, PPV, and NPV could be obtained by varying the nominal threshold (Table 3). A low value (threshold = 1) ensured 100% sensibility and 100% NPV, while a high value (threshold = 5) showed 92.8% specificity and 91.9% PPV.

Those TLE procedures classified as very low risk (T-score = 0) were truly low risk, with no advance extractions occurred. Furthermore, the scoring schema showed an increase in advanced extraction rate with increasing scores, those with a score of 6 had 96.3% advanced extraction rate. Figure 5 depicts advanced extraction, complete success, and complications rates with their 95% CIs according to score class.

Discussion

Main findings

In this article, we developed and validated a novel risk stratification schema to predict the need for advanced tools during TLE using data from two national referral sites. The MB score is a point-based scoring system which can be simply calculated in the preparatory phase of the procedure including the number, the age and the type of targeted leads. We observed a clear increase in advanced procedure risk with an increasing score and a strong predictive value of this schema.

Multiparametric scores for lead extraction

Although registries demonstrated that TLE is a safe and effective procedure,^3,4,9 there is a great interest in developing scoring system to assess the risk of complications or predict the difficulty of the extraction. This awareness may facilitate procedure planning, which includes choice between open surgical and hybrid approach or transfer to a specialized centre, and improve resources management.

Major complications associated with TLE primarily arise from damage to the venous system or myocardium. Their incidence is low ranging from 0.2% to 1.8%,¹⁰ but death if often the result. In our registry, we confirmed a low incidence (0.6%) of major complications with two fatal adverse events. As the incidence of these events is so low, it appears difficult to develop an effective predictor. Brunner et al.³ using a registry of 2999 TLE procedures, identified several variables associated with the risk of life-threating complications: cerebrovascular disease, ejection fraction ≤15%, lower platelet count, international normalized ratio ≥1.2, mechanical sheaths, and powered sheaths. A risk stratification schema to predict complication was also proposed by Fu et al.¹¹ using a single-site database of 702 procedures, they found that major complications rate was basically associated with longer lead duration. High-risk patients (with a  10-year-old pacing or a >5-year-old ICD lead) had significantly higher major events than moderate-risk (with pacing lead 1–10 years old or ICD lead 1–5 years old) and low-risk (any lead ≤1-year-old) patients (5.3%, 1.2%, and 0%, respectively). This schema showed also to be safe and feasible in guiding the selection of operating room vs. device laboratory for lead extraction in a recent report.¹²

Other researches proposed scores to predict 1-year mortality of patients underwent TLE procedures.^13,14 The IKAR score was developed in a study population of 130 patients with a 1-year mortality of 28%. The proposed score was composed by four variables: infective indications, kidney dysfunction, age >56 years, and removal of ICD lead. The patients with the IKAR score = 0 had a rate of 0% of mortality during the follow-up, while the IKAR score >or = 4 was associated with an elevated rate of 1-year mortality (94%).¹³

Difficulty of a lead extraction procedure, rather than complications or all-cause mortality, was the objective of other reports. On the basis of the systematic implementation of a stepwise approach, a first strategy was the prediction for advanced techniques to achieve clinical success. In a first single-centre experience, four clinical and technical risk factors were identified: younger patient age, longer duration of the initial implantation, the number of extracted leads, and the presence of ICD lead. The absence of all the four characteristics was accompanied by 0% PPV for the requirement of powered sheaths for TLE, whereas the coexistence of all four risk factors was characterized by 87% requirement of advanced lead extraction.¹⁵ In a later single-centre experience, the fluoroscopy time of the procedure was taken as an index of difficulty.¹⁶ The lead extraction difficult score was defined as: number of extracted leads within a procedure + lead age (years from implant) + 1 if dual-coil − 1 if vegetation. A subsequent study tested the score in an independent cohort showing its capability to predict complex cases with a sensitivity of 86.9%, a specificity of 70.0%, and an NPV of 98.0%.¹⁷

In the present study, we tried to develop a novel risk stratification schema for TLE to overcome the limitations of the previous published scores. First, we implemented a joint collaboration to create a large multicentre database which could be randomized in derivation and validation cohorts. Second, we prospectively applied a standard stepwise approach which made possible to identify the most difficult procedures considering the need of advanced tools to achieve procedural success. In the validation cohort, the derived MB score showed a strong predictive value with an AUC of 0.80. Low values of the score efficiently identified simple procedures with a rate of advanced extraction of 0% (95% CI 0–26.5%) and an NPV of 100% in case of MB score = 0, and of 30.9% (95% CI 17.6–47.1%) and 75.9% in case of MB score = 1. On the other hand, the PPV reached 91.9% with a nominal threshold of 5, identifying a subgroup of procedures with an advanced extraction rate of 96.3%.

Clinical application

In case of lead dysfunction or infection, choosing the best lead management strategy is crucial and warrants a careful assessment of patients’ characteristics. Extraction should be offered when alternative lead management options appear less favourable to the patient’s immediate and long-term risks¹ and this score may be included in this assessment. But, even when extraction is preferred for long-term benefits, there are some open points which could be discussed and may depend on characteristics of the targeted leads. In the ELECTRa registry, 52% of the procedures were performed in an operating room, 38.5% in an electrophysiological laboratory, and 9.5% in a hybrid room.⁴ In addition, despite major complications seem related to the individual experience of the operator, in clinical practice, as reported by a wide survey conducted by the EHRA, more than one-third of sites treat less than 15 patients per year and 26% from 15 to 30 patients per year.¹⁸

In this scenario, a validate risk stratification score may help the lead management strategies, the logistical approach to lead extraction and have impact on patient safety, resource optimization and costs. The MB score includes only characteristics of targeted leads and can be easily calculated in the preparatory phase of the procedure providing a value ranging from 0 to 6. Extraction procedures with MB score = 0, which are basically patients with one targeted pacing lead implanted from less than 3 years with active fixation, were always extracted by simple traction with or without locking styles (NPV 100%). These patients are a truly low risk and may be safely extracted in low-volume centres. Even patients with MB score = 1 have a low risk to have a complex procedure (NPV 75.9%). The risk of multiple procedures should be avoided; however, for some of these patients who could have difficulties in moving to specialized centres, it may warrant an attempt of TLE in low-volume centres before applying a conservative approach. Furthermore, in case of infection, when there is a need for urgent action an attempt at extraction could be made in a referring centre. This cohort includes patients with two pacing leads actively fixed implanted from less than 3 years, or with one recent (<3 years) ICD lead or one recent pacing lead passively fixed.

Conversely, patients with MB score = 5 or 6, where the highest contribution is provided by the presence of leads implanted from more than 10 years, will likely need advanced tools and their extraction should be managed by high-volume operators with working experience with all techniques and ready to potential complications.

Although all sites which perform TLE procedures should have availability of a wide variety of extraction tools, this simple schema could be applied in multicentre networks to help smaller centres which cannot be proficient in all advanced techniques to identify patients who may be safely and with high procedural success rate treated onsite.

In addition, the MB score could provide additional useful information during the patient informed consent step before the procedure. Tools to simplify the patient evaluation of the risk of the procedure during the review of alternatives to extraction and potentially life-threating complications are needed. Shared decision-making is an important part of current clinical practice and this score might be a mechanism to enable a personalized approach to patient-care.

Limitations

The main limitation of this study is that data derived from two high-volume sites and we cannot exclude bias in patients, as an overestimation of the proportion of complex procedures, and extraction techniques selection. Therefore, extrapolation to centres with less experience could not be straightforward. Further and prospective testing of the MB score in a larger number of sites would add value to this stratification schema.

The proposed score has a strictly technical nature predicting the use of specific extraction tools and was not validated to predict the occurrence of acute complications and the need for surgical intervention. However, it has been reported that the use of powered sheaths itself is a predictor of procedure-related major complications,³ suggesting a potential association of this stratification schema also with procedure-related adverse events.

Conclusions

In this study, we developed and tested in an independent cohort a point-based scoring system to predict the need for advanced tools (powered sheaths or snares by femoral route) to achieve TLE success using a stepwise approach. The MB score, which includes the number, the age, and the type of targeted leads, had a 2.40 increased risk of complex procedure for unitary increase and efficiently identified low- and high-risk patients.

Conflict of interest: D.G. and A.G. are employees of BIOTRONIK Italia. All the remaining authors have no major conflicts of interest to disclose.

References