Body surface activation mapping of electrical ...

Journal of Electrocardiology 51 (2018) 534–541

Contents lists available at ScienceDirect

Journal of Electrocardiology journal homepage: www.jecgonline.com

Body surface activation mapping of electrical dyssynchrony in cardiac resynchronization therapy patients: Potential for optimization☆,☆☆ Alan J. Bank, MD a,⁎, Ryan M. Gage, MS a, Antonia E. Curtin, MS b, Kevin V. Burns, PhD a, Jeffrey M. Gillberg, MS c, Subham Ghosh, PhD c a b c

United Heart & Vascular Clinic, Research Dept., St. Paul, MN, USA The University of Minnesota, Department of Biomedical Engineering, Minneapolis, MN, USA Medtronic, PLC, CRHF, Mounds View, MN, USA

a r t i c l e

i n f o

Keywords: Cardiac resynchronization therapy Body surface mapping Electrical dyssynchrony Optimization Heart failure

a b s t r a c t Background: Electrical synchronization is likely improved by cardiac resynchronization therapy (CRT), but is difficult to quantify with 12-lead ECG. We aimed to quantify changes in electrical synchrony and potential for optimization with CRT using a body-surface activation mapping (BSAM) system. Methods: Standard deviation of activation times (SDAT) was calculated in 94 patients using BSAM at baseline CRT (CRTbl), native, and different CRT configurations. Results: SDAT decreased 20% from native to CRTbl (p b 0.01) and an additional 26% (p b 0.01) at optimal CRT (CRTopt), the minimal SDAT setting. Patients with LBBB and patients with QRS duration ≥150 ms had higher native SDAT and greater decrease with CRTbl (p b 0.01); however, the improvement from CRTbl to CRTopt was similar in all four groups (range: 24–28%). CRTopt was achieved with biventricular pacing in 52% and LV-only pacing in 44%. We propose that improved wavefront fusion demonstrated by BSAMs contributed substantially to the improved electrical synchrony. Conclusion: Optimization potential is similar regardless of pre-CRT QRS morphology or duration. BSAM could possibly improve CRT response by individualizing device programming to minimize electrical dyssynchrony. © 2017 Elsevier Inc. All rights reserved.

Cardiac resynchronization therapy (CRT) improves quality of life, exercise capacity, left ventricular size and function as well as hospitalization rate and mortality in select patients with heart failure (HF) [1,2]. However, minimal or suboptimal response to CRT continues to be a problem, with ~25–30% of patients considered non-responders to therapy [3]. CRT is thought to treat HF by correcting electrical and/or mechanical dyssynchrony. Thus, reasons for non-response to CRT include: 1) lack of native dyssynchrony, or 2) inadequate correction of the underlying dyssynchrony. The only readily clinically available method for

Abbreviations: AV, atrio-ventricular; BiV, biventricular; BSAM, body surface activation mapping; CHB, complete heart block; CRT, cardiac resynchronization therapy; EF, ejection fraction; HF, heart failure; IVCD, interventricular conduction delay; LBBB, left bundle branch block; LV, left ventricle; LVESV, left ventricular end-systolic volume; NYHA, New York Heart Association; RBBB, right bundle branch block; RV, right ventricle; SDAT, standard deviation of activation times; VV, ventricular-ventricular. ☆ Financial support: This study was funded by a grant from Medtronic ERP 2544, PLC. ☆☆ Declaration of interest: Dr. Bank, Mr. Gage, Ms. Curtin, and Dr. Burns receive research grant support from Medtronic. Mr. Gillberg and Dr. Ghosh are employees of Medtronic. ⁎ Corresponding author at: United Heart & Vascular Clinic, 225 N. Smith Ave, Suite 400, St. Paul, MN 55102, USA. E-mail address: [email protected] (A.J. Bank).

https://doi.org/10.1016/j.jelectrocard.2017.12.004 0022-0736/© 2017 Elsevier Inc. All rights reserved.

measuring electrical dyssynchrony is 12-lead ECG QRS duration (QRSd), and changes in QRSd have not correlated well with response to CRT [4]. Numerous studies have focused on pre-CRT patient characteristics that predict response to CRT [4–7]. Other studies have assessed causes of uncorrected dyssynchrony including presence of myocardial scar [8], poor left ventricular (LV) lead location [9], and sub-optimal device programming [10]. However, a major factor hindering further advances in this field is the lack of a practical, sensitive tool for measuring electrical or mechanical dyssynchrony at different time-points (preCRT, during implant, post-CRT) and at multiple programmed device settings. We developed a novel non-invasive body-surface activation mapping (BSAM) technology for generating isochronal activation maps based on 53 body surface electrodes that can be used in the clinical setting to quickly and reproducibly measure electrical dyssynchrony at multiple CRT settings [11]. We have previously demonstrated that improvements in electrical dyssynchrony as measured by this BSAM technique correlate with improved acute hemodynamic response [12] and 6-month LV remodeling response [11]. In the present study, we utilize this technology post-CRT in the outpatient clinic to study the acute impact of different CRT device timings including different atrio-ventricular

A.J. Bank et al. / Journal of Electrocardiology 51 (2018) 534–541

(AV) and ventricular-ventricular (VV) delays on electrical dyssynchrony with the goal of demonstrating the potential of this approach to individually optimize CRT device programming. Methods Patient population We studied 94 HF patients at least 4 months post-CRT who were clinically stable. All patients had EF ≤ 40%, QRSd ≥ 120 ms, and were NYHA class II – ambulatory class IV HF on optimal medical therapy prior to CRT. Patients were previously managed per our standard postCRT protocol which included assessment of AV synchrony early postCRT using mitral inflow patterns at different AV delays and programming AV delay using the iterative method. Patients with CRT implants after April 2014 (n = 48) also underwent 12-lead ECG-guided CRT optimization with the goal of programming to minimize QRSd and maximize electrical wavefront fusion [13,14]. This study was approved by an Institutional Review Board and all patients gave informed consent. 12-lead ECG and ECG Belt Native (CRT off) ECGs were analyzed for rhythm, PR and QRS intervals, and QRS morphology. QRSd was defined as the widest complex on 12-lead ECG. LBBB, right bundle branch block (RBBB), and interventricular conduction delay (IVCD) were defined using standard definitions as previously described [11]. A 53-electrode ECG belt was used for BSAM. Details of this system have been previously described [11, 12]. Briefly, the ECG belt measures body-surface unipolar ECGs and generates activation maps using 17 anterior and 36 posterior electrodes placed on the upper torso. Isochronal maps of electrical activation are created based on body surface activation time at each electrode (time of steepest negative slope of the unipolar QRS complex) as previously described [15]. The earliest activation of any electrode is defined as time zero and activation times at other electrodes are referenced to this time origin. Software then calculates the standard deviation of the 53 individual activation times (SDAT) as a measure of electrical dyssynchrony. Data for electrodes not making contact with the skin were removed from the dyssynchrony calculation and no mathematical interpolation was used.

535

Statistical analysis All continuous variables were expressed as mean ± standard deviation, and categorical variables as count (percentage), unless otherwise noted. Comparisons of continuous variables between groups and within groups were performed using unpaired and paired Student's t-tests as appropriate. STATA/MP software version 14.2 (StataCorp, College Station, TX) was used for data analysis and a two-sided p-value b0.05 was considered statistically significant. Results Patient population There were 94 patients in the study as described in Table 1. Mean EF was 25.7 ± 7% and QRSd was 159 ± 23 ms prior to CRT. QRS morphology was LBBB in 62% of patients. RV-paced patients (n = 14) had underlying CHB and were thus 100% RV-paced. Patients were on good medical therapy and were studied 1.9 ± 2.7 years after CRT implant. At CRTbl, 77 (82%) of patients were programmed to BiV pacing and 17 (18%) to LVonly pacing. Mean AV delay was 129 ± 33 ms (range 60–220 ms). Of the 77 patients BiV-paced at baseline, 48 (62%) had simultaneous LV and RV pacing, 26 (34%) had LV pre-excitation (32 ± 14 ms) and 3 (4%) had RV pre-excitation (23 ± 8 ms). Reproducibility of SDAT and relationship between Δ SDAT and Δ QRS duration Beat-to-beat reproducibility of SDAT was assessed by comparing 195 paired SDAT measurements from 65 different conditions (native and paced) in 23 patients. The correlation coefficient r was 0.99 with the line of fit running through 0 with a slope near 1. Fig. 1 shows this data in a Bland-Altman plot demonstrating a mean difference between two beats of 0 ms and a 95% confidence interval of ±2.5 ms. We have previously shown that reproducibility of SDAT measurements obtained about 30 min apart at the same device settings are also reproducible with a 95% confidence interval of about ±6 ms [11]. We assessed the relationship between % ΔSDAT and % ΔQRS duration in order to determine if the information on electrical dyssynchrony obtained with the ECG belt differed from that obtained with 12-lead ECG. In 45 patients who had 12-lead ECGs acquired at native rhythm

Study protocol Device interrogation was performed to evaluate each patient's underlying AV conduction. ECG belt data was first collected at different AV delays with both biventricular and LV-only pacing, beginning with a programmed AV delay 50% of PR interval and increasing in 20 ms increments until ventricular sensing occurred. If patients were in atrial fibrillation or complete heart block AV delays were not evaluated. Next, ECG belt data was acquired at different VV offsets with AV delay at the baseline value. This was most often assessed at VV simultaneous and with LV pre-excitation of 20, 40, and 60 ms, although in some patients (e.g. patients with RBBB) RV pre-excitation of 20 ms was studied. The baseline programmed setting (pacing mode, AV delay, VV offset) was then evaluated with different LV pacing vectors in the patients with quadripolar LV leads. Data obtained pacing from different quadripolar leads was not reported here as many patients did not have quadripolar leads in place. Patients were also studied with RV-only pacing and with CRT programmed off (native condition). If the patient had a Medtronic device with adaptive-CRT algorithm turned on (n = 25), then algorithm-based programmed parameters at the time of study were considered the CRTbl setting. Activation maps were acquired at an average of 9 ± 3 settings in order to determine the optimal setting with the lowest SDAT, called CRTopt.

Table 1 Demographic and clinical characteristics of patients (n = 94). Characteristic

Value

Age at 1st CRT (years) Male gender n (%) Pre-CRT EF (%) Pre-CRT QRS duration (ms) Pre-CRT QRS morphology LBBB n (%) RBBB n (%) IVCD n (%) RV paced n (%) Pre-CRT NYHA III classification n (%) Pre-CRT ischemic etiology n (%) Pre-CRT COPD n (%) Pre-CRT chronic AF n (%) Pre-CRT diabetes n (%) Pre-CRT CKD (creatinine ≥1.5 mg/dL) n (%) Pre-CRT beta-blocker use n (%) Pre-CRT ACE-I or ARB use n (%)

69 ± 11 64 (68%) 25.7 ± 7 159 ± 23 58 (62%) 6 (6%) 16 (17%) 14 (15%) 72 (77%) 47 (50%) 12 (13%) 11 (12%) 29 (31%) 17 (18%) 88 (94%) 75 (80%)

CRT = cardiac resynchronization therapy; EF = ejection fraction; LBBB = left bundle branch block; RBBB right bundle branch block; IVCD = interventricular conduction delay; RV = right ventricular; NYHA = New York Heart Association; COPD = chronic obstructive pulmonary disease; AF = atrial fibrillation; CKD = chronic kidney disease; ACE-I = angiotensin converting enzyme inhibitor; ARB = angiotensin receptor blocker.

536


Fig. 1. Beat-to-beat reproducibility of SDAT. Bland-Altman plot shows a high degree of reproducibility of SDAT within a single recording of ECG belt data, with 95% confidence interval ± 2.5 ms.

and CRTbl there was no significant correlation between the two electrical dyssynchrony metrics (p = 0.14) as shown in Fig. 2. Electrical dyssynchrony under native, CRTbl and CRTopt settings Fig. 3 shows activation maps from 3 patients at native, CRTbl and CRTopt settings. The first (left) map in each pair shows activation times from the anterior leads (predominantly representing the RV) and the second (right) map in each pair shows activation times from the posterior leads (predominantly representing the LV). In the patient with LBBB, CRTbl setting was LV-only at an AV delay of 90 ms. Shortening the AV delay to 80 ms resulted in complete early activation of the posterior map (LV) and reduction in SDAT from 29 to 17 ms. In the patient with RBBB, CRTbl setting of BiV pacing at an AV delay of 120 ms resulted in an SDAT of 26 ms which improved to 19 ms with increase in AV delay to 160 ms. In the third patient (underlying CHB and RV pacing as preCRT setting), simultaneous BiV pacing improved SDAT from 47 to 24 ms. Preactivating the LV lead by 20 ms at the same AV delay resulted in earlier activation of the LV (posterior map) and further reduction of SDAT to 16 ms. Fig. 4 shows SDAT at native, CRTbl, and CRTopt settings for all patients (A), patients with LBBB vs without LBBB (B), and patients with QRSd

Fig. 2. Relationship between % ΔSDAT and % Δ QRS duration. The correlation in 45 patients in whom 12-lead ECGs were acquired at native rhythm and CRTbl was not significant (p = 0.14).

≥ 150 ms vs 120–149 ms (C). In the overall patient group, SDAT significantly (p b 0.01) decreased by 20 ± 26% from native (41 ± 12 ms) to CRTbl (32 ± 11 ms). SDAT significantly (p b 0.01) decreased an additional 26 ± 21% from CRTbl to CRTopt (SDAT 23 ± 8 ms). In all four subgroups, SDAT significantly decreased from native to CRTbl and from CRTbl to CRTopt except for changes from native to CRTbl in non-LBBB patients. Patients with LBBB had a significantly higher native SDAT than non-LBBB patients (41 ± 10 vs 29 ± 7 ms, p b 0.01) and a significantly greater decrease in SDAT at CRTbl (27 ± 22 vs 0 ± 31%, p b 0.01) but had similar improvement from CRTbl to CRTopt settings (27 ± 23 vs 28 ± 16%, p = 0.80). Patients with QRSd ≥ 150 ms had a significantly higher native SDAT than those with QRSd between 120 and 149 ms (43 ± 11 vs 33 ± 9 ms, p b 0.01) but there was no difference between the groups in improvement from native to CRTbl (23 ± 27 vs 14 ± 27%, p = 0.18) or in optimization potential (27 ± 23 vs 24 ± 17%, p = 0.49). In 48 patients with 12-lead ECG data, pre-CRT and approximately 6 months post-CRT (with CRT programmed off), we did not find any patients with significant electrical reverse remodeling (decrease in QRSd ≥ 15 ms). Baseline and optimal settings for CRT Of the 77 patients BiV-paced at baseline, optimal electrical synchrony was achieved via BiV pacing in 43 (56%), LV-only pacing in 30 (39%), and RV-only pacing in 1 (1%). In 3 (4%) patients, no pacing mode was better than native. Optimal electrical synchrony was achieved by increasing AV delay from CRTbl by 21 ± 52 ms (p b 0.01). Of the 43 patients with both baseline and optimal electrical synchrony achieved via BiV pacing, optimal VV delay was unchanged in 17 (40%), shifted toward more LV pre-activation (34 ± 21 ms) in 17 (40%) and shifted toward less LV pre-activation (33 ± 11 ms) in 9 (20%), as compared to presenting VV delay. In 18 patients with CHB, 65% had optimal electrical dyssynchrony with the LV pre-activated. In addition, 34% of patients without CHB achieved optimal BiV-paced electrical synchrony with LV preactivation of 10 to 60 ms. Of the 17 patients with LV-only pacing at baseline, optimal electrical synchrony was achieved with BiV pacing in 6 (35%) and LV-only pacing in 11 (65%). In the entire group of patients, optimal electrical synchrony was achieved with BiV pacing in 49 (52%) patients, LV-only pacing in 41 (44%), native in 3 (3%) and RV-only in 1 (1%). ECG belt activation maps and waveform fusion ECG belt activation maps provide information on electrical wavefront fusion that can be helpful in understanding the mechanism(s) contributing to improved electrical dyssynchrony under different pacing conditions and provide a physiologic approach to optimizing CRT programming. Fig. 5A shows activation maps from a 57 year old female with an EF of 20%, LBBB, QRSd 140 ms, and PR interval 148 ms. Beneath each map is a diagram depicting the electrical wavefront fusion felt to be occurring at different CRT settings. BiV pacing at an AV delay of 140 ms changed the posterior map from deep blue to a lighter blue (shorter posterior ventricular activation times) and reduced SDAT mildly. BiV pacing at an AV delay close to native PR interval resulted in most of the LV being depolarized by native conduction and only a small contribution from the LV-paced wavefront. The RV-paced wavefront was not contributing to depolarization because LV-only pacing and BiV pacing at an AV delay of 140 ms were identical. As AV delay was decreased while VV delay remained constant, changes in the activation maps occurred due to varying degrees of fusion between LV-paced wavefronts and RV and/or native wavefronts. As the LV was preactivated (row 3) while keeping AV delay constant at 120 ms, a greater proportion of the posterior map (LV) was activated early (red). Thus with the LV paced 50 ms ahead of the RV, SDAT was markedly reduced and almost all of the posterior map was red. The optimal CRT setting was LV-only pacing at a short AV delay of 80 ms. SDAT was very low and activation


537

Fig. 3. Changes in SDAT with application of CRT and optimal CRT. Activation maps with CRT off and CRT at baseline (CRTbl) and optimal (CRTopt) settings for 3 patients with different underlying QRS morphology. Each condition is represented by a pair of anterior (left) and posterior (right) maps. The space between the anterior-posterior maps is the left axilla. Changes in AV or VV delay in each case improve the activation maps and reduce SDAT (see “electrical dyssynchrony under native, CRTbl and CRTopt settings” section for details). SDAT = standard deviation of activation times.

maps showed near-simultaneous activation anteriorly (RV) and posteriorly (LV). The fact that LV-only pacing at short AV delays (80 and 100 ms) and BiV pacing with the LV lead firing 50 ms ahead of the RV lead produced the most synchronous activation maps suggests that there was a delay in the propagation of the LV wavefront as compared to the right-sided (native or RV) wavefronts that could be due to latency, exit block or slow conduction velocity of the LV wavefront. Also shown (Fig. 5B) are 12-lead ECGs at identical AV delays but at different VV delays, and with LV-only pacing. Despite large changes in activation maps and SDAT across these settings, there were only small changes in 12-lead ECG QRS morphology. Activation maps from a 77 year old female with pre-CRT EF 25%, LBBB, and QRSd 120 ms are shown in Fig. 6. The native map showed significant electrical dyssynchrony consistent with LBBB. Atrial pacing was needed due to sinus bradycardia and the atrial-pace to RV-sense interval was prolonged at 250 ms. Simultaneous BiV pacing at paced AV delays between 120 and 200 ms resulted in improvement in SDAT compared to native with the optimal AV delay being 180 ms, where there was likely triple fusion of LV, RV and native wavefronts, and the largest reduction in SDAT. LV-only pacing or BiV pacing with LV preactivation at short and long AV delays resulted in significant heterogeneity of activation times as measured by higher SDAT values. Fig. 7 shows activation maps from a 67 year old male with pre-CRT EF 23%, LBBB, QRS 160 ms and PR interval 140 ms. Marked delay in posterior (LV) activation was present on the native map consistent with LBBB. RV pacing produced a near-identical activation map pattern and similar SDAT. BiV pacing reduced, but did not fully correct, electrical dyssynchrony and resulted in earlier LV activation as evidenced by the change from dark blue to light blue color on the posterior maps. LVonly pacing produced early LV activation (red on the posterior maps) and reduced electrical dyssynchrony as compared to BiV pacing at any given AV delay. Best electrical synchrony, however, was obtained with BiV pacing and the LV lead activated 20 ms ahead of the RV lead at an AV delay of 120 ms. At this setting, SDAT was reduced 70% from 47 ms native to 14 ms optimal. As the LV was preactivated further to 40 and 60 ms ahead of RV pacing, SDAT increased progressively and anterior map (RV) activation occurred progressively later, producing a RBBBtype activation pattern.

Discussion In this study we describe how a new methodology for generating body-surface activation maps can be used to quantify electrical dyssynchrony and determine individualized CRT settings that produce optimal electrical synchrony. We show that CRT programmed clinically at baseline settings reduced electrical dyssynchrony by 20%. This improvement was greater in LBBB patients but similar in patients with and without QRS ≥ 150 ms. The lack of improvement in non-LBBB patients at standard baseline settings may help explain why these patients have typically not responded as well to CRT in many large multi-center studies. However, an important new insight from our study is that at individualized optimal device settings based on BSAM, there is a further 26% improvement in SDAT as compared to current programming. This improvement is of similar magnitude in the subgroups of patients studied, independent of QRS width or morphology. This suggests that standard techniques for CRT optimization achieve only about half of the possible improvement in electrical synchrony and that optimization of electrical synchrony using BSAM may provide similar benefits in many CRT recipients, regardless of their baseline characteristics or whether they are initial responders or non-responders. Comparison of CRT optimization techniques We have previously demonstrated that our measure of electrical dyssynchrony, SDAT, correlates closely with invasively measured rate of LV pressure rise when assessed during CRT implantation with LV leads in different locations [12]. Additionally, we have shown that SDAT pre-CRT, and change in SDAT post-CRT, are better predictors of LV functional and remodeling response to CRT than are QRS duration, LBBB morphology or changes in QRS duration in both Class 1indicated and non-Class 1- indicated (non-LBBB or QRS 120–149 ms) patients [11]. We extend these observations by demonstrating that the ECG belt activation maps and SDAT data can be used post-CRT to understand wavefront fusion and quantify electrical dyssynchrony at different CRT settings—information that can potentially be used to optimize CRT. The reduction in SDAT at CRTopt as compared to native

538


Fig. 4. Electrical dyssynchrony (SDAT) as measured by ECG belt under native (CRT off), baseline programmed, and optimally programmed conditions. Graphs are shown for all patients (A), patients with and without LBBB (B), and patients with QRS b 150 ms vs ≥ 150 ms (C). Electrical dyssynchrony significantly improved in all subgroups from native to CRTbl, except in the non-LBBB sub-group. Electrical dyssynchrony improved significantly and similarly (by 20–30%) in all subgroups at best setting (CRTopt) as compared to CRTbl. *within-group ΔSDAT p b 0.05 between Native and CRTbl conditions. †within-group ΔSDAT p b 0.05 between CRTbl and CRTopt conditions.

rhythm in this study was 44%. Since we did not program patients to their optimal setting in this study we cannot prove the effect on LV remodeling within this cohort. However, we have previously demonstrated the relationship between ΔSDAT (%) and ΔLVESV (%) in 66 CRT patients programmed to baseline CRT settings (ΔLVESV = − 16.9 +

(0.37 × ΔSDAT)) [11]. Based on this information the predicted improvement in LVESV from a 44% decrease in SDAT would be a 33.2% reduction in LVESV, a clinically important remodeling effect. The role of fusion of 3 electrical wavefronts (native conduction, RV pacing and LV pacing) in creating electrical synchrony has been described in detail by Vernooy et al. in studies in dogs [16]. They demonstrate a methodology for achieving optimal pump function by studying various combinations of AV and VV delays and assessing the degree of wavefront fusion on ECG. In addition, Strik, et al. (using contact mapping electrocardiography) showed in dogs with chronic LBBB that optimal hemodynamic effect depended on optimal interaction between LV and/or RV paced wavefronts and intrinsic activation waves as measured by % reduction in total activation time [17]. Similar to these findings in dogs, we found in humans that several different AV and VV combinations can provide similar (and possibly optimal) resynchronization and that LV preactivation is frequently beneficial and necessary in order to achieve optimal electrical synchrony. Device-based algorithms have been developed to try to automatically optimize CRT pacing parameters based on intracardiac timing measurements. Studies comparing use of these algorithms with standard optimization techniques (primarily echocardiography) have only shown non-inferiority [18,19]. The “gold standard” for measuring electrical synchrony in humans is non-invasive electrocardiographic imaging (ECGi) using a N 200 electrode vest, a CT scan of the heart, and complex computer analysis to provide detailed information on epicardial activation [20,21]. Similar to ECGi, our technique records anterior and posterior torso body surface electrograms and utilizes the steepest slope of electrograms to assess local activation; however, our BSM methodology lacks the detail and spatial resolution of ECGi. Despite these limitations, advantages of our technique over ECGi for routine clinical care include: ease of use, lower cost, and no need for CT scans (and the associated contrast). Multiple other non-invasive methods have been used to assist in optimization of CRT. These include echocardiographic measurement of aortic or mitral flow velocities [22], tissue Doppler imaging [23,24], 3D echocardiography [25], noninvasive cardiac output measurement [26], and finger photoplethysmography [27] to name a few. Despite the availability of these various techniques, only a small percentage of CRT patients undergo optimization and these patients are typically nonresponders [28]. Reasons for this include issues related to time, training, expertise, reproducibility, sensitivity to detect small changes, precision, and cost of optimization, as well as the lack of consensus on the best methodology. In addition, many of these methods do not directly address reduction of electrical dyssynchrony, which is an important aspect of cardiac resynchronization. For example, echocardiography-based methods of AV optimization target diastolic LV filling, usually showing a range of AV delays that provide adequate LV filling, but neglect the effect of AV delay on ventricular electrical synchrony. Usually pacing at a short AV delay results in cell to cell conduction of RV and LV paced wavefronts with minimal involvement of Purkinje conduction and minimal contribution from the native electrical wavefront. In many patients, this is sub-optimal and longer AV delays allow fusion of the LV-paced wavefront with the native and/or RV-paced wavefronts (Fig. 7). However, in patients with either RBBB, mildly or moderately prolonged QRSd (b 150 ms), or significant LV wavefront propagation delay (Fig. 5), the AV delay needs to be short to allow the LV wavefront to fuse with the native wavefront and contribute substantially to resynchronization. If the AV delay is not short enough, then the native and/or RV-paced right-sided electrical wavefronts precede the LV wavefront and depolarize most, or all, of the LV. This scenario cannot usually be diagnosed via routine device check as the LV lead will be capturing at a normal threshold—it simply has less non-refractory myocardium to depolarize. VV optimization is not performed routinely in CRT patients, mainly due to lack of a method that can reliably reflect and quantify its effect.


539

Fig. 5. ECG belt maps and depictions of wavefront fusion. A 57 year old female with ischemic cardiomyopathy (see “ECG belt activation maps and waveform fusion” section for details) was studied with ECG belt. Panel A: Colored arrows represent electrical wavefronts traversing the LV from native conduction, RV lead, LV lead, and combined native/RV lead. Electrical dyssynchrony is mildly improved with BiV pacing at all AV delays. LV pre-activation of 40 or 50 ms at constant AV delay improves synchrony. LV-only pacing improves electrical synchrony in a step-wise fashion as AV delay is shortened. ECGs (Panel B) from the same patient under different conditions demonstrate the insensitivity of the ECG for detecting changes in electrical activation and dyssynchrony shown using the ECG belt.

But it may have an important role. In our study, 12 of 18 patients (66%) with CHB had optimal electrical synchrony with LV preactivation. In addition, about 1/3 of patients without CHB had optimal electrical synchrony with LV preactivation. In many CRT patients, the LV myocardial substrate is diseased leading to slow and impaired conduction in the LV. Additionally, the LV lead is epicardial and there may be delay in transmitting the electrical signal from the coronary sinus vein to the mid-myocardium or endocardium. Pre-exciting the LV provides more time for the LV-paced wavefront to depolarize and propagate across the LV, achieving better overall electrical synchrony. Studies using 12lead ECG markers of wavefront fusion (such as increased R waves in V1/V2 and narrower QRSd) to optimize VV timing have shown benefit [13,14]. We used this 12-lead ECG optimization method in about half of our patients as part of our standard clinical CRT optimization protocol. While this resulted in improved electrical synchrony at baseline settings compared to native, the use of the ECG belt substantially increased the improvement in electrical synchrony.

The BSAM technique described in this paper thus offers a number of advantages over current methods for attempting CRT optimization. It is relatively fast, inexpensive, and easy to incorporate within the clinical workflow of device follow-up. The maps and SDAT values are reproducible [11] and there is minimal observer bias since the information is processed by a computer, usually without need for operator interference or modification. BiV vs LV-only pacing Forty-four percent of patients achieved optimal electrical synchrony with LV-only pacing. This finding is consistent with other studies using invasive hemodynamics and echocardiography. Recent studies have shown that device algorithms employing LV pacing synchronized to RV activation result in improved clinical outcomes, including a reduced risk of atrial fibrillation [29], a decreased risk of death or HF hospitalization [30], and a decrease in 30 day hospital readmissions [31]. In

540


Fig. 6. ECG belt maps in patient with moderate (QRS 120 ms) pre-CRT electrical dyssynchrony. Body surface activation maps in a 77 year old female with hypertrophic cardiomyopathy (see “ECG belt activation maps and waveform fusion” section for details). There is significant heterogeneity of activation times on the posterior map at all LV-only setting and at all settings with LV pre-activation. With simultaneous BiV pacing, longer AV delays result in lower SDAT likely due to fusion of predominantly native and LV-paced wavefronts.

contrast, other multicenter randomized studies have not demonstrated a benefit to LV-only pacing over BiV pacing [32]. A limitation of these studies is that CRT was not individualized. In some patients BiV pacing

is better, in some patients LV-only pacing is better, and in many patients either mode can be used if programmed correctly. ECG belt technology helps determine which patients electrically resynchronize optimally

Fig. 7. Body surface activation maps in a 67 year old male with ischemic cardiomyopathy. Native and RV pacing maps show marked activation delay of the posterior (predominantly LV) map. Electrical dyssynchrony is improved mildly at all BiV settings with simultaneous ventricular activation. LV-only pacing also improves electrical dyssynchrony, but BiV pacing with the LV lead pre-activated by 20 ms produces optimal electrical synchrony.


with which pacing configuration. In our study, 39% of patients presenting with BiV pacing had optimal electrical synchrony with LV-only pacing and 35% presenting with LV-only pacing were optimally paced BiV. Limitations This was a retrospective study performed post-implant. The patients were studied using their clinically programmed LV pacing electrodes for an average of 9 ± 3 device settings. We did not evaluate every possible combination of AV delay, VV delay and/or pacing vector. Although many patients had quadripolar LV leads, we did not report the effects of changing LV pacing electrode on electrical synchrony because some patients did not have quadripolar leads and we were limited in the number of CRT settings we could study. Lastly, this was an observational study where we acutely evaluated the impact of device programming changes on electrical dyssynchrony but did not make any permanent changes in programming based on BSAM data. Future studies are required to show how BSAM-guided programming affects LV remodeling/function and clinical outcomes. Conclusion Current clinical programming of CRT devices only achieves about half of the reduction in electrical dyssynchrony possible. BSAM can noninvasively quantify electrical dyssynchrony at multiple device settings and identify the setting in an individual patient that provides the lowest electrical dyssynchrony. Our technique provides a practical, clinicallyapplicable approach to CRT programming that offers the possibility of improving CRT response.

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

[19]

Role of the funding source This study was funded by an External Research Program grant from Medtronic, PLC. Medtronic loaned the ECG belt hardware and software for the purposes of this study. All authors developed the concept/design of the study. Data was collected and analyzed by Mr. Gage, Ms. Curtin, and Dr. Ghosh. Data was interpreted by all authors, with statistical analysis performed by Mr. Gage. Dr. Bank and Mr. Gage drafted the article, with critical revision and final approval of the article provided by all authors. Data statement

[20]

[21]

[22]

[23]

Our data is unavailable to access and unsuitable to post, as the ECG belt is still under development and the research data is confidential.

[24]

References

[25]

[1] Bristow MR, Saxon LA, Boehmer J, Krueger S, Kass DA, De Marco T, et al. Cardiacresynchronization therapy with or without an implantable defibrillator in advanced chronic heart failure. N Engl J Med 2004;350:2140–50. [2] Cleland JG, Daubert JC, Erdmann E, Freemantle N, Gras D, Kappenberger L, et al. The effect of cardiac resynchronization on morbidity and mortality in heart failure. N Engl J Med 2005;352:1539–49. [3] Birnie DH, Tang AS. The problem of non-response to cardiac resynchronization therapy. Curr Opin Cardiol 2006;21:20–6. [4] Gold MR, Thebault C, Linde C, Abraham WT, Gerritse B, Ghio S, et al. Effect of QRS duration and morphology on cardiac resynchronization therapy outcomes in mild heart failure: results from the resynchronization reverses remodeling in systolic left ventricular dysfunction (REVERSE) study. Circulation 2012;126:822–9. [5] Bank AJ, Gage RM, Marek JJ, Onishi T, Burns KV, Schwartzman D, et al. Mechanical dyssynchrony is additive to ECG criteria and independently associated with reverse remodelling and clinical response to cardiac resynchronisation therapy in patients with advanced heart failure. Open Heart 2015;2:e000246. [6] Bilchick KC, Kamath S, DiMarco JP, Stukenborg GJ. Bundle-branch block morphology and other predictors of outcome after cardiac resynchronization therapy in Medicare patients. Circulation 2010;122:2022–30. [7] Dupont M, Rickard J, Baranowski B, Varma N, Dresing T, Gabi A, et al. Differential response to cardiac resynchronization therapy and clinical outcomes according to QRS morphology and QRS duration. J Am Coll Cardiol 2012;60:592–8. [8] Bilchick KC, Dimaano V, Wu KC, Helm RH, Weiss RG, Lima JA, et al. Cardiac magnetic resonance assessment of dyssynchrony and myocardial scar predicts function class

[26]

[27]

[28]

[29]

[30]

[31]

[32]

541

improvement following cardiac resynchronization therapy. J Am Coll Cardiol Cardiovasc Imaging 2008;1:561–8. Singh JP, Klein HU, Huang DT, Reek S, Kuniss M, Quesada A, et al. Left ventricular lead position and clinical outcome in the multicenter automatic defibrillator implantation trial-cardiac resynchronization therapy (MADIT-CRT) trial. Circulation 2011; 123:1159–66. Mullens W, Grimm RA, Verga T, Dresing T, Starling RC, Wilkoff BL, et al. Insights from a cardiac resynchronization optimization clinic as part of a heart failure disease management program. J Am Coll Cardiol 2009;53:765–73. Gage RM, Curtin AE, Burns KV, Ghosh S, Gillberg JM, Bank AJ. Changes in electrical dyssynchrony by body surface mapping predict left ventricular remodeling in cardiac resynchronization therapy patients. Heart Rhythm 2017;14:392–9. Johnson WB, Vatterott PJ, Peterson MA, Bagwe S, Underwood RD, Bank AJ, et al. Body surface mapping using an ECG belt to characterize electrical heterogeneity for different left ventricular pacing sites during cardiac resynchronization - relationship with acute hemodynamic improvement. Heart Rhythm 2017;14:385–91. Cooper JM, Patel RK, Emmi A, Wang Y, Kirkpatrick JN. RV-only pacing can produce a Q wave in lead 1 and an R wave in V1: implications for biventricular pacing. Pacing Clin Electrophysiol 2014;37:585–90. Sweeney MO, Hellkamp AS, van Bommel RJ, Schalij MJ, Borleffs CJ, Bax JJ. QRS fusion complex analysis using wave interference to predict reverse remodeling during cardiac resynchronization therapy. Heart Rhythm 2014;11:806–13. Akahoshi M, Hirai M, Inden Y, Sano H, Shimizu A, Kondo T, et al. Body-surface distribution of changes in activation-recovery intervals before and after catheter ablation in patients with Wolff-Parkinson-White syndrome: clinical evidence for ventricular ‘electrical remodeling’ with prolongation of action-potential duration over a preexcited area. Circulation 1997;96:1566–74. Vernooy K, Verbeek XA, Cornelussen RN, Dijkman B, Crijns HJ, Arts T, et al. Calculation of effective VV interval facilitates optimization of AV delay and VV interval in cardiac resynchronization therapy. Heart Rhythm 2007;4:75–82. Strik M, van Middendorp LB, Houthuizen P, Ploux S, van Hunnik A, Kuiper M, et al. Interplay of electrical wavefronts as determinant of the response to cardiac resynchronization therapy in dyssynchronous canine hearts. Circ Arrhythm Electrophysiol 2013;6:924–31. Ellenbogen KA, Gold MR, Meyer TE, Fernndez Lozano I, Mittal S, Waggoner AD, et al. Primary results from the SmartDelay determined AV optimization: a comparison to other AV delay methods used in cardiac resynchronization therapy (SMART-AV) trialcular delay programming in cardiac resynchronization therapy. Circulation 2010;122:2660–8. Martin DO, Lemke B, Birnie D, Krum H, Lee KL, Aonuma K, et al. Investigation of a novel algorithm for synchronized left-ventricular pacing and ambulatory optimization of cardiac resynchronization therapy: results of the adaptive CRT trial. Heart Rhythm 2012;9:1807–14. Ploux S, Eschalier R, Whinnett ZI, Lumens J, Derval N, Sacher F, et al. Electrical dyssynchrony induced by biventricular pacing: implications for patient selection and therapy improvement. Heart Rhythm 2015;12:782–91. Ploux S, Lumens J, Whinnett Z, Montaudon M, Strom M, Ramanathan C, et al. Noninvasive electrocardiographic mapping to improve patient selection for cardiac resynchronization therapy: beyond QRS duration and left bundle branch block morphology. J Am Coll Cardiol 2013;61:2435–43. Jansen AH, Bracke FA, van Dantzig JM, Meijer A, van der Voort PH, Aarnoudse W, et al. Correlation of echo-Doppler optimization of atrioventricular delay in cardiac resynchronization therapy with invasive hemodynamics in patients with heart failure secondary to ischemic or idiopathic dilated cardiomyopathy. Am J Cardiol 2006;97:552–7. Bank AJ, Burns KV, Kelley AS, Thelen AM, Kaufman CL, Adler SW. Echocardiographic improvements wtih pacemaker optimization in the chronic post cardiac resynchronization therapy setting. Clin Med Cardiol 2008;2:65–74. Gage RM, Burns KV, Vatterott DB, Kubo SH, Bank AJ. Pacemaker optimization in nonresponders to cardiac resynchronization therapy: left ventricular pacing as an available option. Pacing Clin Electrophysiol 2012;35:685–94. Sonne C, Bott-Flugel L, Hauck S, Hadamitzky M, Lesevic H, Demetz G, et al. Threedimensional echocardiographic optimization improves outcome in cardiac resynchronization therapy compared to ECG optimization: a randomized comparison. Pacing Clin Electrophysiol 2014;37:312–20. Khan FZ, Virdee MS, Pugh PJ, Read PA, Fynn SP, Dutka DP. Non-invasive cardiac output measurements based on bioreactance for optimization of atrio- and interventricular delays. Europace 2009;11:1666–74. Whinnett ZI, Davies JE, Willson K, Chow AW, Foale RA, Davies DW, et al. Determination of optimal atrioventricular delay for cardiac resynchronization therapy using acute non-invasive blood pressure. Europace 2006;8:358–66. Gras D, Gupta MS, Boulogne E, Guzzo L, Abraham WT. Optimization of AV and VV delays in the real-world CRT patient population: an international survey on current clinical practice. Pacing Clin Electrophysiol 2009;32(Suppl. 1):S236–9. Birnie D, Hudnall H, Lemke B, Aonuma K, Lai-Fun Lee K, Gasparini M, et al. Continuous optimization of cardiac resynchronization therapy reduces atrial fibrillation in heart failure patients: Results of the adaptive cardiac resynchronization therapy trial. Heart Rhythm 2017;14:1820–5. Birnie D, Lemke B, Aonuma K, Krum H, Lee KL, Gasparini M, et al. Clinical outcomes with synchronized left ventricular pacing: analysis of the adaptive CRT trial. Heart Rhythm 2013;10:1368–74. Starling RC, Krum H, Bril S, Tsintzos SI, Rogers T, Hudnall JH, et al. Impact of a novel adaptive optimization algorithm on 30-day readmissions: evidence from the adaptive CRT trial. J Am Coll Cardiol Heart Fail 2015;3:565–72. Thibault B, Ducharme A, Harel F, White M, O'Meara E, Guertin MC, et al. Left ventricular versus simultaneous biventricular pacing in patients with heart failure and a QRS complex ≥120 ms. Circulation 2011;124:2874–81.