Author + information
- Received June 12, 2009
- Revision received December 1, 2009
- Accepted December 7, 2009
- Published online June 1, 2010.
- Niraj Varma, MA, DM⁎⁎ (, )
- Ping Jia, PhD†,
- Charulatha Ramanathan, PhD† and
- Yoram Rudy, PhD‡
- ↵⁎Reprint requests and correspondence:
Dr. Niraj Varma, Cardiac Electrophysiology, Cleveland Clinic, 9500 Euclid Avenue, Cleveland, Ohio 44195
Objectives To compare right ventricular (RV) activation during intrinsic conduction or pacing in heart failure (HF) patients.
Background RV activation during intrinsic conduction or pacing in patients with left ventricular (LV) dysfunction is unclear but may affect the prognosis. In cardiac resynchronization therapy (CRT), timed LV pacing (CRT-LV) may be superior to biventricular pacing (CRT-BiV), and is hypothesized to be due to the merging of LV-paced and right bundle branch–mediated wavefronts, thus avoiding perturbation of RV electrical activation.
Methods Epicardial RV activation duration (RVAD) (onset to end of free wall activation) was evaluated noninvasively by electrocardiographic imaging in healthy control subjects (n = 7) and compared with that of HF patients (LV ejection fraction 23 ± 10%, n = 14). RVAD in HF was contrasted during RV pacing, CRT-BiV, and CRT-LV at optimized AV intervals.
Results During intrinsic conduction in HF (n = 12), the durations of QRS and precordial lead rS complexes were 158 ± 24 and 77 ± 17 ms, respectively, indicating delayed total ventricular depolarization but rapid initial myocardial activation. Echocardiography demonstrated no significant RV disease. RV epicardial voltage, activation patterns, and RVAD in HF did not differ from normal (RVAD 32 ± 15 vs. 28 ± 3 ms, respectively, p = 0.42). In HF, RV pacing generated variable areas of slow conduction and prolonged RVAD (78 ± 33 ms, p < 0.001). RVAD remained delayed during CRT-BiV at optimized atrioventricular intervals (76 ± 32 ms, p = 0.87). In contrast, CRT-LV reduced RVAD to 40 ± 26 ms (p < 0.016), comparable to intrinsic conduction (p = 0.39) but not when atrioventricular conduction was poor or absent.
Conclusions In HF patients without RV dysfunction treated with CRT, normal RV free wall activation in intrinsic rhythm indicated normal right bundle branch–mediated depolarization. However, the RV was vulnerable to the development of activation delays during RV pacing, whether alone or with CRT-BiV. These were avoided by CRT-LV in patients with normal atrioventricular conduction.
- cardiac resynchronization therapy
- electrocardiographic imaging
- heart failure
- left bundle branch block
- right ventricle
- right ventricular pacing
Right ventricular (RV) electrical activation during intrinsic conduction or ventricular pacing is not well characterized. This understanding may influence cardiac resynchronization therapy (CRT) mode in heart failure (HF) patients. For example, left ventricular (LV) pacing only (CRT-LV) may be superior to simultaneous biventricular pacing (CRT-BiV) (1). However, the postulated mechanism of timed LV-paced wavefronts merging with intrinsic conduction, thus avoiding any RV pacing effect, has not been directly tested. The explanation assumes intact right bundle branch (RBB) conduction in patients with both Purkinje system and ventricular disease and also that the RV pacing component of CRT-BiV exerts deleterious effects. Certainly, RV pacing alone increases patient morbidity (2), which can be attributed to detrimental electromechanical effects simulating left bundle branch block (LBBB) (3). Interestingly, RV pacing increases mortality in HF patients with pre-existing LBBB (4), indicating contributory mechanisms other than altered LV activation alone. These may include pacing-induced alteration of RV activation and ensuing RV contractile dysfunction, which diminishes survival in HF (5–7).
Our preliminary reports indicated that RV epicardial breakthrough in HF patients was delayed compared with that of the healthy control subjects and right ventricular activation duration (RVAD) prolonged by RV pacing (8,9). Here, we extended these observations. We hypothesized that RV pacing as part of CRT-BiV alters RV activation compared with intrinsic RBB-mediated depolarization and may be avoided by CRT-LV. This was tested using electrocardiographic imaging (ECGI), which noninvasively depicts epicardial cardiac excitation with high resolution (9,10).
Seven normal, healthy adults (age 21 to 43 years, 4 men) with normal electrocardiograms formed a control group (data published previously ) to enable comparison with baseline RV activation in HF. Fourteen HF patients (age 61 ± 18 years, 11 men, LV ejection fraction <35%) who received CRT were studied 6.8 ± 5 months after implantation. In 12 of 14 patients with preserved atrioventricular (AV) conduction, surface electrocardiograms were analyzed for QRS complex durations and rS intervals in V1/V2 (11) and voltage maps generated during intrinsic conduction. ECGI reconstructs epicardial voltages assuming a homogeneous torso without taking into account the conductivities of tissues surrounding the heart. This facilitates practical application of ECGI in patients without compromising accurate reconstruction of voltage patterns (e.g., low-voltage regions). Although absolute voltage values may not be preserved, relative magnitudes (ratios of potential amplitudes in different regions) are reconstructed under the homogeneous torso approximation (12,13). This methodology permitted accurate evaluation of epicardial RV voltage relative to LV values.
In the study group, RV activation times and patterns were contrasted during intrinsic conduction and in different pacing modes. RV activation duration (RVAD) was defined as the time difference between the onset and end of RV free wall depolarization. Atrial synchronous pacing was performed in all patients except 1 with atrial fibrillation. The AV interval was optimized (AVopt) echocardiographically during CRT-BiV according to Ritter et al. (14). RV pacing effects were evaluated in dual-chamber RV pacing with shortened AV interval (30 to 50 ms), i.e., when ventricular excitation was committed to the paced wavefront. Dual-chamber RV pacing was then repeated at AVopt to permit direct comparison with CRT. CRT was assessed in CRT-BiV and CRT-LV at AVopt when AV conduction was intact. Implanted devices varied in ability to perform RV and LV pacing independently. Thus, both modes were not obtained in every patient.
Data are reported as mean ± SD. Groups were compared using unpaired 2-tailed t tests (SPSS software, version 13, SPSS Inc., Chicago, Illinois). A value of p < 0.05 was considered significant.
HF patients (ischemic etiology in 50%) had a poor New York Heart Association functional class and reduced LV function (ejection fraction 23 ± 9.8%) (Table 1). Echocardiographically, the RV size was normal in 13 of 14 patients but enlarged in Patient #12. Two of 14 patients without intrinsic conduction had received CRT upgrades from long-term RV pacing (QRS >200 ms). Twelve of 14 patients exhibited intrinsic conduction (PR interval 203 ± 53, range 140 to 350 ms), including 2 with QRS interval <120 ms selected on echocardiographic criteria. AVopt was shorter (143 ± 31, range 80 to 189 ms, p < 0.001). Ten patients conducted with LBBB (QRS 158 ± 24, range 130 to 194 ms). The rS duration in V1/V2 was 77 ± 17 ms, indicating rapid initial myocardial activation (Fig. 1A). RV epicardial free wall voltage was preserved in all HF patients, including Patient #12, although LV low-voltage areas were evident (Fig. 1B).
RV activation patterns in HF patients (including Patient #12) (Table 2) were similar to those in the normal, healthy control subjects (although these were not age or sex matched) (Figs. 1C and 2A). RV epicardial breakthrough(s) followed QRS onset. Sites varied similarly in both groups. In HF, RV breakthrough occurred at various anterolateral free wall positions in 11 of 12 patients. In Patient #10, the earliest breakthrough was inferoposteriorly followed by an independent lateral RV free wall breakthrough 20 ms later. Epicardial activation spread radially from breakthrough sites. The latest RV activation occurred basally. No areas of slow conduction were demonstrated. The duration of the entire RV free wall activation in HF patients did not differ from that in the healthy control subjects (32 ± 15 vs. 28 ± 3 ms , respectively, p = 0.42) (Fig. 2B).
This was performed apically except in patients 1, 6, and 14, in whom leads were placed mid-septally. RV pacing (dual-chamber RV pacing) with minimal AV delay prolonged RV activation by slowing conduction, indicated by crowded isochrones not observed during intrinsic conduction (Fig. 3). Overall, RVAD was 83 ± 26 ms compared with 32 ± 15 ms during intrinsic conduction (p < 0.001) (Fig. 4). RV mid-septal pacing did not differ. When AV intervals were extended during RV pacing, RVAD remained prolonged (78 ± 33 ms) and unchanged compared with pacing at short AV intervals (p = 0.70). In Patients #3 and #13, RV pacing did not generate local conduction slowing, and the RVAD was <45 ms for both short and long AV intervals, similar to intrinsic conduction. This was maintained during CRT-BiV (see the CRT section).
RVAD during CRT-BiV was 76 ± 32 ms. This did not differ from RV pacing in either short (p = 0.56) or extended (p = 0.87) AV delays, (Figs. 3 and 4). In 4 patients with intrinsic conduction, RV epicardial breakthrough sites observed during intrinsic conduction were retained, indicating manifestation of RBB conduction during CRT-BiV (Fig. 3). The effect of this on RV activation was usually negligible, and the overall RVAD remained prolonged compared with intrinsic conduction. Again, Patients #3 and #13 were exceptions. In these patients, RVAD during CRT-BiV was as rapid as that during RV pacing with extended AV delays or during intrinsic conduction (i.e., RV pacing did not generate local slowing and instead contributed to RV activation) (Fig. 5).
During CRT-LV in patients with preserved AV conduction (n = 10), the RV was activated more rapidly (40 ± 26 ms) compared with either RV pacing or CRT-BiV (p < 0.016) (Figs. 3 and 4). RV excitation was unaccompanied by regions of conduction slowing, as observed during RV pacing or CRT-BiV. Overall, RVAD was similar to intrinsic conduction (40 ± 26 ms vs. 32 ± 15 ms, p = 0.39). During CRT-LV, RV breakthrough sites and activation patterns were identical to intrinsic conduction; that is, determined by intrinsic RBB conduction (Fig. 3), except in Patient #4. In this patient, the AVopt (130 ms) was considerably shorter than the PR interval (350 ms). Thus, LV pacing did not permit intrinsic conduction, and prolonged RVAD (91 ms vs. 25 ms during intrinsic conduction) indicated slow depolarization from LV pacing without RBB effect. When intrinsic conduction was absent (Patients #2 and #6), RV activation was similarly committed to the LV-paced wavefront. RVAD determined by LV pacing without RBB participation in Patients #2, #4, and #6 was 79 ± 15 ms, which was indistinguishable from RV pacing (78 ± 33 ms).
Survival in HF patients is reduced with LBBB and RV pacing, but improved by CRT (2,15,16). Thus, electrical activation sequences affect the prognosis. However, human cardiac activation studies under these conditions have usually reported LV effects only (17). RV activation has been rarely studied, although RV dysfunction further increases mortality in HF patients (6,7). Normally, RV contraction is a complicated peristaltic movement beginning in the inflow region and extending to the outflow tract (18). Altered electrical activation with RV pacing may perturb RV hemodynamics (1). However, electrical characterization poses technical difficulties. The RV is largely silent during conventional electrocardiography because it generates weak electrical forces completed early in the QRS complex and mostly concealed by LV depolarization. However, ECGI reveals RV free wall electrical activation in detail.
In the HF patients studied, echocardiography indicated minimal RV disease. Voltage maps revealed no regions with diminished potential, in contrast to the LV (Fig. 1). In patients with LBBB (or narrow QRS configuration), surface electrocardiographic recordings demonstrated rapid initial myocardial activation (short rS duration) suggestive of intact RBB conduction (11) despite the presence of LV conduction abnormalities (8). ECGI activation maps supported this: after delayed RV breakthrough, both the pattern and speed of RV epicardial free wall activation were similar to those of the healthy control subjects (Fig. 2). (In contrast, the same patients exhibited LV conduction delay/block during intrinsic activation.) After this, there was radial and rapid centrifugal spread of activation across the RV free wall (Fig. 2). These results in HF patients are consistent with previous observations (8) and similar to findings in normal hearts (9,19). The activation pattern likely reflects the course of the RBB, which passes down the septum to the base of the anterior papillary muscle and then fans out into multiple free-running false tendons terminating in the free wall as a profuse subendocardial Purkinje network. This generates nearly simultaneous activation of the free wall in a radial manner, likely responsible for initiation of RV contraction from the inflow tract to outflow tract (20,21). In contrast, the LV free wall depolarizes from apex to base. RV activation duration in this study was determined from completion of free wall depolarization. Previous invasive endocardial mapping in similar patients revealed lateral wall activation times nearly identical to those obtained here with ECGI (22).
The electrical effects of RV pacing are conceived as reproducing those of LBBB. In support, QRS configurations are similar and LV activation occurs transseptally after RV depolarization in both. For RV pacing to truly simulate LBBB, it should replicate RV activation by the RBB. However, ECGI, which discloses electrical activation not evident on conventional electrocardiography, demonstrated that RV pacing significantly altered the pattern and velocity of RV depolarization. RV-paced wavefronts propagated slowly from apex to base, in contrast to rapid and radial spread during intrinsic activation. RVAD was prolonged almost 3-fold. Activation maps revealed conduction slowing, with isochronal crowding around pacing sites (Fig. 3). This indicates development of functional delays, presumably due to predominant cell-to-cell propagation, with limited or no engagement of intact Purkinje conduction system tissue. This is similar to the mechanism of LV activation during RV pacing (3,23). When AV delay was extended in patients with preserved AV conduction, RV depolarization was still initiated by RV apical pacing, but intrinsic RBB conduction also manifested itself. RV activation then was determined by the sum effect of these wavefronts. Overall, there was no change in RVAD compared with RV pacing at shorter AV delays because pacing effects dominated, although in 2 patients (Patients #3 and #13), RV pacing generated comparatively less functional free wall conduction slowing and RVAD remained <45 ms, comparable to intrinsic conduction (Fig. 5).
The results illustrate that prolongation of duration of global RV activation by RV pacing was driven by slow conduction areas generated locally around the stimulus site. These were usually small in extent. Delay permitted intrinsic RBB-mediated conduction to contribute to RV free wall depolarization, resulting in varying degrees of wavefront fusion. RV pacing may also disturb septal depolarization, but this is not depicted by epicardial mapping. Hence, the pattern and duration of RV free wall activation were the outcome of the balance of intrinsic (centrifugal) and RV-paced (centripetal) wavefronts. When RV-paced delays were less, global RV activation duration was not delayed (although direction of depolarization was different from intrinsic conduction). In contrast, when intrinsic AV conduction was absent or poor, the RV was committed to activation by the RV pacing and RVAD was longer.
RVAD and pattern were unaltered by CRT-BiV compared with the effects of RV pacing alone at the same AV delay. Thus, RV depolarization in CRT-BiV was governed by RV electrode excitation (Fig. 3). Overall, CRT-BiV slowed RV activation while simultaneously causing LV pre-excitation. LV pacing with AVopt (CRT-LV with preserved AV conduction) produced strikingly different effects on RV activation compared with those of CRT-BiV or RV pacing. In the majority, CRT-LV permitted RV activation via intact RBB with centrifugal RV activation, avoiding functional delays generated by RV stimulation. Swifter RV activation was reflected by shorter overall RVAD, comparable to intrinsic conduction in the same patients (Figs. 4 and 5). In 3 patients (#3, #4, and #13), CRT-LV did not decrease the RVAD compared with CRT-BiV. In 2 patients (Patients #3 and #13), the RVAD during CRT-LV and CRT-BiV was short (<45 ms) because RV-paced wavefronts synergized with intrinsic RBB conduction and maintained a normal RVAD. In Patient #4, the RVAD remained delayed compared with intrinsic conduction because AVopt was considerably shorter than the intrinsic PR interval (130 ms vs. 350 ms, respectively), preventing emergence of RBB-mediated activation. In this case, and in those with absent intrinsic conduction (Patients #2 and #6), when RV activation resulted from LV pacing, RVAD was 79 ± 15 ms, indistinguishable from RV pacing (78 ± 33 ms), suggesting a propagating mechanism of slow cell-to-cell conduction.
Here, in patients with poor LV function and preserved AV conduction treated with CRT, ECGI demonstrated intact RBB conduction and electrically normal RV function despite LBBB. In these patients, intrinsic rapid and confluent RV free wall activation could be retained with CRT-LV, avoiding RV delays generated by RV pacing, either alone or as part of CRT-BiV.
These data provide a mechanistic explanation for previous assumptions that avoidance of RV pacing by timed LV pacing may have a hemodynamic benefit (1). RV pacing–induced changes in activation sequence and prolongation of activation duration may all potentially perturb the normal sequential pattern of RV inflow-to-outflow contraction. However, the current study also demonstrated that improvement of RV activation by CRT-LV cannot be expected to occur consistently, even in those with intact RBB conduction, as observed in 3 of 10 patients. Thus, CRT-LV did not further improve RV activation when RVAD remained normal during CRT-BiV or RV pacing (e.g., Patient #3 [Fig. 5]), or when the paced AV interval was much shorter than the intrinsic PR interval (Patient #4) and the RV was committed to electrode stimulation in any case. AV optimization is an important feature of correct post-implantation programming (24), and the influence of the AV interval on the contribution of intrinsic conduction to ventricular activation was illustrated previously (8). The current study illustrates that any incremental electrical benefit to be gained by CRT-LV in terms of RV activation (observed in 7 of 10 patients) depends on the balance between intrinsic and paced AV intervals and effects of functional conduction slowing with pacing, which are unpredictable.
The study in an unselected CRT population is limited by the small number of subjects, and thus findings may not be characteristic of all CRT patients (e.g., RV activation may differ in patients with RV dysfunction ). Examination was performed at variable time intervals after implantation (Table 2), and a remodeling effect that influenced the results cannot be excluded. CRT-LV advantage with respect to RV activation, suggested by the current study, may be diminished if pacing cannot be timed with RBB conduction (i.e., changing AV intervals with activity or irregular conduction in atrial fibrillation) or if RBB conduction is absent. The chronic effects of pacing on RV function may be complex and the ultimate CRT response likely determined by both LV (17) and RV effects. For example, during long-term CRT-BiV, RV mechanical dysfunction improvement may occur by LV remodeling (26) or by RV septal site pacing (25).
Although CRT-LV was not inferior to CRT-BiV in 1 trial (27), the significant variations reported here suggest that demonstration of CRT-LV benefit requires attention to individual electrical substrate. Device programming may be guided noninvasively by dynamic 3-dimensional mapping, such as ECGI, and merits prospective evaluation.
This study was supported by NIH-NHLBI Merit AwardR37-HL-033343 and grant RO1-HL-49054 to Dr. Rudy. Dr. Rudy co-chairs the scientific advisory board and holds equity in CardioInsight Technologies (CIT). CIT does not support any research conducted by Dr. Rudy, including that presented here. Drs. Jia and Ramanathan are equity holders and paid employees of CIT. Dr. Varma is not an equity holder, advisory board member, or employee of CIT. Patients were drawn from Dr. Varma's practice. CIT provided institutional grant support to Loyola University for case studies 9 through 14. Data for case studies 1 through 8 were processed by Dr. Jia while a PhD student in Dr. Rudy's laboratory at Case Western Reserve University and contributed to Dr. Jia's PhD thesis.
- Abbreviations and Acronyms
- optimized atrioventricular interval
- cardiac resynchronization therapy
- biventricular pacing
- left ventricular pacing
- electrocardiographic imaging
- heart failure
- left bundle branch block
- left ventricular/ventricle
- right bundle branch
- right ventricular/ventricle
- right ventricular activation duration
- Received June 12, 2009.
- Revision received December 1, 2009.
- Accepted December 7, 2009.
- American College of Cardiology Foundation
- Hayes J.J.,
- Sharma A.D.,
- Love J.C.,
- et al.
- Ghio S.,
- Gavazzi A.,
- Campana C.,
- et al.
- Di Salvo T.G.,
- Mathier M.,
- Semigran M.J.,
- Dec G.W.
- Ramanathan C.,
- Jia P.,
- Ghanem R.,
- Ryu K.,
- Rudy Y.
- Ritter P.,
- Padeletti L.,
- Gillio-Meina L.,
- Gaggini G.
- Durrer D.,
- van Dam R.T.,
- Freud G.E.,
- et al.
- Myerburg R.J.,
- Nilsson K.,
- Gelband H.
- Fantoni C.,
- Kawabata M.,
- Massaro R.,
- et al.
- Prinzen F.,
- Spinelli J.C.,
- Auricchio A.
- Auricchio A.,
- Stellbrink C.,
- Block M.,
- et al.
- Bleeker G.B.,
- Schalij M.J.,
- Nihoyannopoulos P.,
- et al.
- Rao R.K.,
- Kumar U.N.,
- Schafer J.,
- et al.