Author + information
- Received October 23, 2015
- Revision received January 14, 2016
- Accepted February 18, 2016
- Published online August 1, 2016.
- Mintu P. Turakhia, MD, MSa,∗ (, )
- Michael Cao, MDb,
- Avi Fischer, MDc,
- Yelena Nabutovsky, MSc,
- Laurence S. Sloman, BSc,
- Nirav Dalal, MSc and
- Michael R. Gold, MD, PhDd
- aDivision of Cardiology, Stanford University, Palo Alto, California
- bGolden Heart Medical, Rosemead, California
- cSt. Jude Medical, Sylmar, California
- dDivision of Cardiology, Medical University of South Carolina, Charleston, South Carolina
- ↵∗Reprint requests and correspondence:
Dr. Mintu Turakhia, Veterans Affairs Palo Alto Health Care System, Stanford University School of Medicine, 3801 Miranda Avenue, 111C, Palo Alto, California 94304.
Objectives The study sought to compare survival, lead deactivation, and lead replacement with quadripolar versus bipolar leads using a retrospective cohort of patients with newly implanted cardiac resynchronization therapy (CRT) systems.
Background In CRT, quadripolar left ventricular (LV) leads offer alternative pacing sites and vectors not available with bipolar LV leads, which may improve the effectiveness of the therapy.
Methods Using nationwide data from device implant registration records of a single manufacturer, we identified patients with a de novo cardiac resynchronization therapy with defibrillation (CRT-D) implanted between November 30, 2011, and May 31, 2013. Patients were followed for up to 24 months. The primary predictor was LV lead type (quadripolar Quartet [St. Jude Medical, St. Paul, Minnesota] LV lead or bipolar LV lead). The primary outcome was death and the secondary outcomes were LV lead replacement and deactivation.
Results Among 23,570 patients (69.5 ± 11.1 years of age; 28% female; median follow-up time 1.14 years), 18,406 had quadripolar and 5,164 had bipolar LV leads. The quadripolar and bipolar groups had 5.04 and 6.45 deaths per 100 patient-years, respectively (p < 0.001). After multivariate adjustment, the quadripolar lead was associated with a lower risk of deactivation (hazard ratio [HR]: 0.62; 95% confidence interval [CI]: 0.46 to 0.84; p = 0.002), replacement (HR: 0.67; 95% CI: 0.55 to 0.83; p < 0.001), and death (HR: 0.77; 95% CI: 0.69 to 0.86; p < 0.001).
Conclusions In this observational study of CRT-D devices, use of a quadripolar, compared to a bipolar LV lead, was associated with a reduction in LV lead deactivation, replacement, and mortality.
Although cardiac resynchronization therapy (CRT) has been shown to decrease the risk of death and other heart failure events (1–4), its effectiveness can be limited by lead migration or dislodgement, high ventricular capture thresholds, phrenic nerve capture, and clinical nonresponse. These findings have been well characterized in clinical trials and observational studies of unipolar and bipolar transvenous left ventricular (LV) leads. In 2011, a quadripolar LV lead and CRT system was approved in the U.S. market (St. Jude Medical Inc., Sylmar, California). This lead, which has 4 electrodes, allows for LV pacing from 10 unique pacing vectors, the majority of which are not available in bipolar CRT systems, thereby offering more options for vector configuration. Although single-center studies, small multicenter studies, and 1 randomized trial have found a lower prevalence or risk of short-term complications compared to unipolar and bipolar lead data (5–10), there are very limited real-world data comparing long-term outcomes by lead type. We therefore sought to address this evidence gap by evaluating the comparative effectiveness of quadripolar versus bipolar CRT for clinical outcomes including lead deactivation, lead replacement, and survival.
We performed a nationwide retrospective cohort study using linked data from 4 sources: device implant registration, device remote monitoring, postal (zone improvement plan [ZIP]) code sociodemographic data, and Social Security Death Index. We obtained data without patient identifiers from implant registration records of all CRT devices manufactured by St. Jude Medical, Inc. Data included date of implant, age at implant, sex, patient ZIP code, site ZIP code, and device lead and model numbers. To evaluate outcomes of LV lead deactivation, we obtained data for the subset of patients enrolled in Merlin.net remote monitoring, including enrollment date, transmission date, and device configuration and programming at each transmission.
Because baseline characteristics were limited to data from device enrollment records, we supplemented the device dataset with regional sociodemographic information matched on the basis of the patient’s residence. We used ZIP code–level data on median income and college education from the 2007 to 2011 5-year summary of the American Community Survey (ACS). The ACS is an ongoing, mandatory statistical survey, administered by the U.S. Census Bureau, that samples a small percentage of the population every year to plan government services (11). Patient-level ZIP code was mapped to the ACS ZIP Code Tabulation Area, and regional characteristics were assigned at the patient level. This method of linking patient characteristics to ZIP codes for studying clinical data was described previously (12,13). After linking to regional characteristics, the analysis cohort was geographically de-identified by removing ZIP and ZIP Code Tabulation Area codes.
We obtained death information from the Social Security Death Index Master File, which included all death records through November 30, 2013, the defined administrative censoring date. The Social Security Death Index has been shown to be an accurate tool for determining the mortality status (14). In addition, death reports through this date made directly to St. Jude Medical device registration by health care providers or family members were used, although this accounted for a very small proportion (<1%) of ascertained deaths.
We included patients, 18 to 89 years of age, who were implanted with St. Jude Medical CRT with defibrillation (CRT-D) pulse generators in the United States between November 30, 2011, and May 31, 2013. This time period was used because it corresponds to the time after introduction of the quadripolar lead for clinical use in the United States (during which time bipolar leads continued to be available), thereby minimizing the potential for secular trends if nonoverlapping observation periods were used across LV lead types. All enrolled patients were followed until November 30, 2013. The models included were Unify (3231-40 and 3231-40Q), Unify Assura (3257-40 and 3257-40Q), Unify Quadra (3249-40 and 3249-40Q), and Quadra Assura (3265-40 and 3265-40Q). To minimize the potential for confounding and bias, we limited the cohort to de novo CRT patients by first excluding patients who had received any St. Jude Medical CRT pulse generator or LV lead prior to November 30, 2011. Patients were excluded who were not documented as having received an LV lead on the same day as the included CRT-D pulse generator, as it was suspected that these patients had a previously implanted LV lead and were receiving a replacement pulse generator. Lastly, patients were excluded who did not have gender information available. The final cohort was separated into 2 groups: patients who received a quadripolar system (Unify Quadra and Quadra Assura) and patients who received a bipolar system (Unify and Unify Assura).
The primary outcome was time to death from any cause. Patients who did not have a record of death were assumed to be alive and censored on the administrative censoring date or the date of lead replacement. We evaluated differences in mortality with and without adjustment for age, sex, remote monitoring enrollment, and socioeconomic factors of household income, and percent of population with ≥4 years of college education.
Potential confounders of differences in the primary outcome were evaluated. First, we performed stratified analyses of the primary outcome on the basis of the proportion of high versus low biventricular (BiV) pacing, because differences in BiV pacing could mediate survival outcomes. However, %BiV pacing data, which relies on ascertainment of device diagnostics in follow-up, were only available in patients with remote monitoring. Therefore, this secondary outcome was evaluated only in the subset of patients who were enrolled in remote monitoring, had at least 1 transmission in the 365 days after implant, and had %BiV pacing data available for at least 70% of their follow-up period. A cutpoint of %BiV pacing ≥93% was used, as this was shown previously to be associated with the greatest magnitude of benefit from CRT therapy (15).
Next, we evaluated for potential mediators of outcome differences, by comparing secondary outcomes of time to LV lead replacement and time to LV lead deactivation. LV lead replacements and corresponding date were ascertained from St. Jude Medical device registration records. Patients were censored on the date of death, date of lead replacement, or the administrative censoring date. Differences in replacement were evaluated with and without adjustment for age, sex, remote monitoring enrollment, household income, and percent of population with ≥4 years of college education. Deactivation data also relies on ascertainment of device configuration and programming changes in follow-up, so were only available in patients with remote monitoring. Therefore, this secondary outcome was only evaluated in the subset of patients who were enrolled in remote monitoring and had at least 1 transmission in the 365 days after implant. The date of LV lead deactivation was identified via the first transmission in which LV pacing was disabled. Because almost all of the deactivations occurred in the first year after implant, these data were censored at 365 days post-implant, at the date of death, date of lead replacement, or at the administrative censoring date. Differences in deactivation were evaluated with and without adjustment for age, sex, household income, and percent of population with ≥4 years of college education.
Patient characteristics and outcomes were compared between the quadripolar and bipolar groups. Continuous variables, including age, regional median income, and regional college attendance were compared using a Student t test or Mann-Whitney test if the distribution was not normal. Categorical variables, including sex and enrollment in remote monitoring were compared using a chi-square test. Mortality in the 2 groups was calculated using incidence rates (deaths per 100 patient-years) and compared using a chi-square test. For all outcomes, cumulative event-free survival was measured by the Kaplan-Meier method, and differences were compared using the log-rank test and Cox proportional hazards models. The proportional hazards assumption was tested using Schoenfeld residuals and was met. For multivariate analysis, multivariate Cox regression was performed with covariates selected on the basis of baseline differences. Interaction testing was used for subgroup analyses. For the secondary outcomes of LV lead replacement and LV lead deactivation, a Cox model accounting for the competing risk of death was used (16). All calculations were performed in R, version 3.0.2, augmented with the following R packages: survival (17), epiR (18), cmprsk (19), and stats (20).
There were 23,570 patients who met the inclusion criteria, 18,406 in the quadripolar group and 5,164 in the bipolar group (Figure 1). There were no differences between groups in age or sex (Table 1). Remote monitoring enrollment was slightly higher in the quadripolar group (49% vs. 46%; p < 0.001) (Table 1). There were small but statistically significant differences in the regional median household income and the regional proportion with a college degree (Table 1). In addition, more than 96% of implants were performed at sites within ZIP codes implanting both quadripolar and bipolar leads. Early in the observation window, more bipolar devices were implanted than quadripolar; subsequently, this pattern reversed. Thus, the median follow-up time was lower for the quadripolar (1.06 years; interquartile range: 0.72 to 1.46) than the bipolar group (1.52 years; interquartile range: 0.96 to 1.82; p < 0.001). Of the final quadripolar and bipolar groups, 8,915 and 2,367 patients, respectively, had at least 365 days of remote monitoring and were used to evaluate the secondary outcome of LV lead deactivation. A further subset of 7,086 quadripolar and 1,887 bipolar patients had %BiV pacing data available for at least 70% of their follow-up period and was used for stratified analysis of mortality.
There were 1,463 deaths over 27,065 patient-years of follow-up. The quadripolar lead group had a significantly lower incidence rate of death compared to the bipolar lead group (5.04 deaths vs. 6.45 deaths per 100 patient-years; p < 0.001) (Table 2). In univariate analysis, the quadripolar lead implantation was significantly associated with a decreased risk of death (hazard ratio [HR]: 0.74; 95% confidence interval [CI]: 0.66 to 0.82; p < 0.001) (Table 3). Remote monitoring was also associated with improved survival (HR: 0.42; 95% CI: 0.38 to 0.47; p < 0.001), whereas age and male sex increased risk of death (Table 3). Unadjusted Kaplan-Meier survival curves show separation of the confidence intervals between both groups between 90 and 120 days (Figure 2). After adjustment for age, sex, remote monitoring enrollment, household income, and percent of population with ≥4 years of college education, the quadripolar lead remained significantly associated with reduced risk of death (HR: 0.77; 95% CI: 0.69 to 0.86; p < 0.001) (Table 3).
In stratified analysis of 8,973 patients with remote %BiV pacing data, the proportion of patients with high %BiV pacing did not differ between those with quadripolar and bipolar leads (86.6% and 86.9%, respectively; p = 0.83). The quadripolar lead remained associated with survival in both high and low %BiV pacing groups (high %BiV pacing: HR: 0.77; 95% CI: 0.60 to 0.99; p = 0.039; low %BiV pacing: HR: 0.56; 95% CI: 0.36 to 0.85; p = 0.007) without significant interaction (p = 0.29 for interaction).
The incidence of LV lead deactivation and replacement was lower in quadripolar patients (1.96 vs. 3.18 per 100 patient-years, p < 0.001; 1.47 vs. 1.90 per 100 patient-years, p = 0.015, respectively) (Table 2). In the presence of a competing risk of mortality, the quadripolar lead was associated with a lower risk of deactivation (HR: 0.62; 95% CI: 0.46 to 0.84; p = 0.002) (Figure 3) and replacement (HR: 0.67; 95% CI: 0.55 to 0.83; p < 0.001) (Figure 4) after adjustment for covariates.
Of 8,915 patients with quadripolar leads with at least 365 days of remote monitoring, 4,143 (46.5%) were programmed to a quadripolar-exclusive vector, 4,677 (52.5%) to a traditional vector, and 95 (1%) were deactivated at the start of their follow-up period. Between implant and final follow-up, a total of 529 (5.9%) patients were reprogrammed. Of 257 patients reprogrammed from an initial quadripolar-exclusive vector, 241 switched to a traditional vector and 16 were deactivated. Of 245 patients reprogrammed from a traditional vector, 213 went to a quadripolar-exclusive vector and 32 were deactivated. There were also 27 patients who switched from an initial deactivated state to a quadripolar-exclusive vector (9 patients) or a traditional vector (18 patients).
In a large, real-world population of U.S. patients with newly implanted CRT-D devices, use of the quadripolar LV lead, compared to a bipolar lead, was associated with a lower risk of death, LV lead replacement, and LV lead deactivation, even after adjustment for patient characteristics. Findings were consistent in patients with high and low %BiV pacing. Taken together, these findings may indicate greater effectiveness of CRT associated with quadripolar leads.
A recent observational study of 721 patients from 3 centers in the United Kingdom compared lead performance and mortality between patients implanted with a quadripolar and a bipolar lead over a 5-year period (21). In contrast to our study, the UK study included patients with bipolar LV systems from all manufacturers. The majority of the quadripolar systems (96%) were from the same manufacturer as the present study. The study found a similar risk reduction for mortality (adjusted HR: 0.65; 95% CI: 0.46 to 0.96; p = 0.03) and of LV lead revisions. Phrenic nerve stimulation, which we were unable to directly ascertain as an outcome, was completely eliminated with reprogramming in 55 patients, compared to in 24 of 40 (60%) in the bipolar group. The present study extends these findings by using complete U.S. data from company device registration, therefore avoiding potential biases of registries or of limited sites. The UK study found a 48% reduction in fluoroscopy dose to the operator, which is noteworthy.
There are several potential mechanisms for improved survival. Previous reports have provided evidence for acute benefits of quadripolar LV systems, including the ability to pace around scar tissue (7) and noninvasive resolution of phrenic nerve stimulation (5,8). Therefore, the benefit may be directly mediated by the observed lower risk of LV lead deactivation, thereby preserving a longer duration of CRT pacing. Second, a reduced risk of LV lead replacement could reduce exposure to potential complications of lead or device revisions, which have been shown to be substantial (22). Third, vector reprogramming, on the basis of preservation of LV capture, minimization of phrenic nerve stimulation, or CRT optimization, may mediate observations. An earlier study showed acute changes in echocardiography septal-to-lateral wall delay and global longitudinal strain delivered from different quadripolar lead configurations (23). We show that 49% of patients had a quadripolar-specific vector programmed at some point during the observation period, although devices were programmed to and from quadripolar-specific vectors at similar proportions. The quadripolar lead may allow pacing at less apical LV sites by pacing from the more proximal electrodes, because apical pacing sites are associated with worse outcomes in CRT (24,25). Importantly, these findings appear independent of the %BiV pacing, suggesting the observed results were not due simply to more LV pacing, but rather more effective pacing. Unfortunately, we are unable to determine the reasons for vector changes or deactivation in our analysis. Ongoing prospective studies of the quadripolar lead in multipoint pacing configurations, in which more than 1 LV bipolar pacing site can be used simultaneously, may help to add clarity (26–28). However, these and other potential mediators require more detailed, granular investigation before definitive mechanistic links can be established.
Findings were also independent of remote monitoring, which itself was associated with increased survival, consistent with prior observations (29). Unfortunately, just under half of patients were enrolled in remote monitoring, and an examination of barriers and facilitators of its use in CRT-D requires exploration.
First, this is an observational analysis where LV lead assignment is not random, leading to concern for unidentified confounding and bias. Selection bias may arise from factors that cannot be evaluated from the available data—such as baseline status, clinical compliance with CRT implant guidelines, and insurance coverage. Because of the data source, our baseline variables were limited to only age, sex, and remote monitoring enrollment. However, because all 3 of these variables have been shown to be tightly associated with survival in heart failure patients with ICD or CRT (29–33), we might expect that signs of covariate imbalance or confounding would be first apparent in these variables. However, age and sex were identical in both groups, and it is generally uncommon that baseline differences are present when age and sex distributions are so similar in large unselected cohorts.
We also found no evidence of a socioeconomic gradient across lead types that could indicate the potential for unidentified confounders affecting patient compliance or LV lead selection. Furthermore, adjustment for remote monitoring, which did demonstrate an association with improved survival, can also indirectly account for these factors, for example if optimal medical therapy or medication compliance were associated with use of remote monitoring. Site or implanter level effects could also explain results if there was a mismatch among quality, performance, or patient selection by sites that implanted one type of lead versus the other. For example, the increased uptake of the quadripolar lead over time could suggest that implanters in the early study period may have been more comfortable with the bipolar lead and therefore more likely to use it in more complex patients. However, 96% of implants were performed at sites within ZIP codes implanting both quadripolar and bipolar leads and geographic location was included in the multivariate analysis.
Importantly, information on the potential confounders of QRS duration and bundle branch block pattern was not available, but it is unlikely that this could account for the present observations. It is well established that patients without left bundle branch block (LBBB) and QRS duration <150 ms derive less benefit from CRT in clinical trials (30,31,34,35) and in observational studies (33). In a pairwise comparison of patients with LBBB and QRS duration ≥150 ms compared to no LBBB and QRS duration of 120 to 149 ms, the HR for mortality is 0.85 (95% CI: 0.94 to 0.77) (33). Thus a complete QRS duration imbalance would be needed in both groups as an unidentified confounder to account for the observed benefit of the group with a quadripolar system (HR: 0.77; 95% CI: 0.69 to 0.86; p < 0.001). Such a drastic selection bias in the quadripolar group toward patients with LBBB and QRS duration ≥150 ms is very unlikely. Finally, median follow-up was between 1 and 2 years per patient. Studies with longer follow-up and an assessment of health care utilization may help to ensure that survival curves do not converge and to provide an assessment of value.
In this large, nationwide, cohort study of patients receiving de novo CRT, we found that use of a quadripolar lead, compared to bipolar lead, was independently associated with improved survival and a reduction in both risk of LV lead deactivation and replacement.
COMPETENCY IN MEDICAL KNOWLEDGE: CRT can improve survival and symptoms in patients with heart failure and LBBB. However, lead location, phrenic nerve stimulation, and high capture thresholds can decrease benefit. Quadripolar LV leads, compared to bipolar leads, are associated with improved survival and decreased risk of replacement and deactivation. Quadripolar pacing LV leads can be considered in patients who meet criteria for CRT. After implantation, ongoing follow-up and vector configuration of the quadripolar lead may be essential to sustain its potential benefits.
TRANSLATIONAL OUTLOOK: Randomized trials or observational analyses that can clarify mediators of improved survival associated with quadripolar LV leads may provide further support of these findings.
The authors would like to thank Edith Arnold for initial analysis tool development and early drafts of the manuscript and Charisma Kumar for extensive help with analysis (Charisma Kumar and Edith Arnold have agreed to include their names in the Acknowledgements).
Dr. Turakhia has served as a consultant for Biotronik, Precision Health Economics, St. Jude Medical, Medtronic, iRhythm, and Gilead Sciences; has received research grants from St. Jude Medical, Medtronic, iRhythm, and Gilead Sciences; and has received lecture honoraria from St. Jude Medical. Dr. Cao has no relationships relevant to the contents of this paper to disclose. Dr. Fisher, Ms. Nabutovsky, Mr. Sloman, and Mr. Dalal are employees of and own stock in St. Jude Medical. Dr. Gold has served as a consultant for Medtronic, St. Jude Medical, and Boston Scientific; and has received research grants from St. Jude Medical and Boston Scientific.
- Abbreviations and Acronyms
- American Community Survey
- confidence interval
- cardiac resynchronization therapy
- cardiac resynchronization therapy with defibrillation
- hazard ratio
- left bundle branch block
- left ventricular
- zone improvement plan
- Received October 23, 2015.
- Revision received January 14, 2016.
- Accepted February 18, 2016.
- American College of Cardiology Foundation
- van Everdingen W.M.,
- Cramer M.J.,
- Doevendans P.A.,
- Meine M.
- Boriani G. More options available with a quadripolar LV lead provide in-clinic solutions to CRT challenges. Available at: https://www.clinicaltrials.gov/ct2/show/results/NCT01510652?sect=X01256#all. Accessed December 15, 2015.
- ↵American Community Survey (ACS). Available at: https://www.census.gov/programs-surveys/acs/. Accessed December 14, 2015.
- Brennan J.M.,
- Peterson E.D.,
- Messenger J.C.,
- et al.
- Koplan B.A.,
- Kaplan A.J.,
- Weiner S.,
- Jones P.W.,
- Seth M.,
- Christman S.A.
- ↵Therneau T. Survival analysis, including penalised likelihood. 2.36–14 ed, 2012.
- ↵Stevenson M. An R package for the analysis of epidemiological data. 0.9–48 ed, 2013.
- ↵Gray RJ. Subdistribution analysis of competing risks. 2.2–7 ed, 2015.
- ↵The R Stats Package. Available at: http://stat.ethz.ch/R-manual/R-patched/library/stats/html/00Index.html. Accessed December 14, 2015.
- Behar J.M.,
- Bostock J.,
- Zhu Li A.P.,
- et al.
- Poole J.E.,
- Gleva M.J.,
- Mela T.,
- et al.
- Singh J.P.,
- Klein H.U.,
- Huang D.T.,
- et al.
- Thebault C.,
- Donal E.,
- Meunier C.,
- et al.
- Leclercq C. MOre REsponse on Cardiac Resynchronization Therapy With MultiPoint Pacing (MORE-CRT MPP). Available at: https://www.clinicaltrials.gov/ct2/show/NCT02006069?term=multipoint+pacing&rank=5. Accessed December 15, 2015.
- Tomassoni G. MultiPoint Pacing IDE Study. Available at: https://www.clinicaltrials.gov/ct2/show/NCT01786993?term=multipoint+pacing&rank=1. Accessed December 14, 2015.
- Saxon L.A.,
- Hayes D.L.,
- Gilliam F.R.,
- et al.
- Goldenberg I.,
- Moss A.J.,
- Hall W.J.,
- et al.
- Zareba W.,
- Klein H.,
- Cygankiewicz I.,
- et al.
- Arshad A.,
- Moss A.J.,
- Foster E.,
- et al.
- Cleland J.G.,
- Abraham W.T.,
- Linde C.,
- et al.
- Gold M.R.,
- Thebault C.,
- Linde C.,
- et al.