Optimized Left Ventricular Endocardial Stimulation Is Superior to Optimized Epicardial Stimulation in Ischemic Patients With Poor Response to Cardiac Resynchronization Therapy

Objectives The purpose of this study was to identify the optimal pacing site for the left ventricular (LV) lead in ischemic patients with poor response to cardiac resynchronization therapy (CRT). Background LV endocardial pacing may offer benefit over conventional CRT in ischemic patients. Methods We performed cardiac magnetic resonance, invasive electroanatomic mapping (EAM), and measured the acute hemodynamic response (AHR) in patients with existing CRT systems. Results In all, 135 epicardial and endocardial pacing sites were tested in 8 patients. Endocardial pacing was superior to epicardial pacing with respect to mean AHR (% change in dP/dtmax vs. baseline) (11.81 [-7.2 to 44.6] vs. 6.55 [-11.0 to 19.7]; p = 0.025). This was associated with a similar first ventricular depolarization (Q-LV) (75 ms [13 to 161 ms] vs. 75 ms [25 to 129 ms]; p = 0.354), shorter stimulation–QRS duration (15 ms [7 to 43 ms] vs. 19 ms [5 to 66 ms]; p = 0.010) and shorter paced QRS duration (149 ms [95 to 218 ms] vs. 171 ms [120 to 235 ms]; p < 0.001). The mean best achievable AHR was higher with endocardial pacing (25.64 ± 14.74% vs. 12.64 ± 6.76%; p = 0.044). Furthermore, AHR was significantly greater pacing the same site endocardially versus epicardially (15.2 ± 10.7% vs. 7.6 ± 6.3%; p = 0.014) with a shorter paced QRS duration (137 ± 22 ms vs. 166 ± 30 ms; p < 0.001) despite a similar Q-LV (70 ± 38 ms vs. 79 ± 34 ms; p = 0.512). Lack of capture due to areas of scar (corroborated by EAM and cardiac magnetic resonance) was associated with a poor AHR. Conclusions In ischemic patients with poor CRT response, biventricular endocardial pacing is superior to epicardial pacing. This may reflect accessibility to sites that cannot be reached via coronary sinus anatomy and/or by access to more rapidly conducting tissue. Furthermore, guidance to the optimal LV pacing site may be aided by modalities such as cardiac magnetic resonance to target delayed activating sites while avoiding scar.

which is not constrained to the epicardial coronary venous anatomy, may provide superior hemodynamics (10)(11)(12) and improved CRT response, which may be of particular benefit in ischemic patients and nonresponders to conventional CRT (12)(13)(14). The site of optimal LVendo stimulation is highly variable in ischemic and nonischemic groups (15) with no reliable method to guide optimal LVendo lead placement. Cardiac magnetic resonance (CMR) could potentially identify the target for LV lead placement, being able to delineate scar and dyssynchrony (7). Endocardial contact mapping can demonstrate exquisite detail of endocardial activation as well as location and size of myocardial scar. Because patients with ischemic cardiomyopathy and myocardial scar have the poorest response to CRT and the most to gain from LVendo pacing, advanced imaging and mapping modalities may be able to guide the optimal site for endocardial LV lead delivery.
We hypothesized that in a group of ischemic patients (with demonstrable myocardial scar on CMR) and a high prevalence of CRT nonresponse, LVendo pacing would produce a superior hemodynamic response compared with the optimal epicardial response (LVepi). Furthermore, by pacing multiple sites, we sought to investigate whether the optimal site of LV stimulation (both epicardially and endocardially) could be predicted on the basis of scar and/or the latest point of electrical activation. By comparing endocardial contact mapping data with CMR, we further sought to elucidate the mechanisms of improved response with LVendo pacing and whether these imaging modalities could be used to guide the optimal LVendo pacing sites.

METHODS
The study complied with the Declaration of Helsinki and the protocol was approved by the local ethics committee. Informed consent was obtained from each patient. Patients with ischemic cardiomyopathy (judged by significant coronary artery disease and myocardial fibrosis on CMR), QRS duration of <150 ms, and previously implanted CRT (mean duration of implant 26 AE 21 months) have a phenotype of suboptimal response to CRT and were intentionally selected for study (3). Baseline assessment before CRT implant included clinical assessment (New York Heart Association functional class), 12-lead electrocardiogram, and 2-dimensional echocardiography. Patients underwent an extensive endocardial mapping protocol and acute hemodynamic study.
CMR data were compared with contact mapping and hemodynamic data findings to compare optimal LVendo and LVepi pacing locations. HEMODYNAMIC   The acute hemodynamic response (AHR) was measured using an 0.014-inch high-fidelity Certus RADI PressureWire in the LV as previously described (11). Atrial pacing 10 bpm greater than the intrinsic rate was used as baseline and compared with con-     Mixed effect model for all data points achieving capture comparing epicardial and endocardial pacing across the dependent variables as shown. A total of 32 epicardial and 87 endocardial data points were compared across 8 patients.
AHR ¼ acute hemodynamic response; QLV ¼ first ventricular depolarization (earliest onset QRS duration on surface 12 lead electrocardiogram) to the nadir signal on the LV lead electrogram.      Table 1).

DISCUSSION
We studied the optimal site for both LVepi and LVendo stimulation in a cohort of patients with ischemic heart disease with poor response to conventional CRT. The principal findings were as follows.
1. Indiscriminate endocardial pacing was superior to epicardial stimulation, associated with a similar first ventricular depolarization, shortened stimulation-QRS duration and shortened paced QRS duration.

Optimal achievable endocardial AHR was superior
to the optimal achievable epicardial AHR. In keeping with the current study, LVendo pacing has been found to be superior to conventional CRT in both ischemic and nonischemic patients (11)(12)(13). Derval et al. (22) found that LVendo pacing was superior to posterolateral LVepi pacing in nonischemic patients with significant individual variation between the optimal LVendo pacing sites (22). Spragg et al. (10) undertook EAM and AHR measurement in 11 patients with ischemic heart disease and dyssynchronous heart failure at the time of ventricular tachycardia ablation and found LVendo was superior to LVepi pacing (10). Our group has shown the superiority of LVendo pacing over conventional CRT in a group of ischemic and nonischemic patients (11). The superiority of LVendo pacing over LVepi seems to be reproducible, but with a significant variability in the optimal site between patients in all the aforementioned studies (10)(11)(12).
The current study has some important differences In keeping with this finding, the Q-LV at the optimal LVendo sites were not significantly longer than those at the optimal LVepi site. It seems likely that, on the basis of our and other prior findings, the latest activating site is not necessarily the optimal site to pace with respect to improved hemodynamics. This may be due to localized areas of slow conduction with islands of viable tissue within areas of scar that activate late (and therefore have a long Q-LV). Likewise, when stimulation is performed at that site, impulse propagation is also slow out of this area and does not result in effective resynchronization. An example of this is seen in Figure 5 in a patient with a large circumferential midventricular and apical infarct with areas of late activating tissue within the scar. It is possible that seeking a late activated site may be beneficial, but only if conduction out of that site is not also delayed or blocked by regions of scar. This is analogous to a Goldilocks effect where the optimal site in ischemic patients because of scar/slow conduction may be not too early, not too late but just right, somewhere in between the two.  Dilated, globular heart with a heavy burden of myocardial scar. Earliest activation is white and latest activation blue/purple. In this case, LVendo locations were not superior to conventional LVepi with respect to the AHR. The point of latest electrical activation in this case is around the anteroseptum, most likely as a result of slow activation spreading and encircling a large region of scar. Although these sites are the latest activated they will not produce a good AHR because they are in scar and may explain why the latest activated site is not always the optimal pacing site. Abbreviations as in Figures 1 and 3. epicardial pacing. Our results support image guidance on the basis that a broad range of AHR values were obtained and therefore not all endocardial positions are equal in this cohort. Although an indiscriminate approach showed endocardial pacing was superior to epicardial pacing ( Table 2), an even greater AHR was achievable when both epicardial and endocardial sites were optimized. Therefore, an image-guided, targeted approach could be a strategy for identifying the optimal location for LV lead stimulation. CMR techniques do, however, require further evaluation to assess their merit in guiding endocardial pacing sites and techniques which allow CMR derived scar and mechanical activation to be fused onto live fluoroscopy for epicardial LV lead guidance may also be used for LVendo lead guidance (18,25).

CONCLUSIONS
Our findings suggest that, in ischemic patients with poor CRT response, endocardial pacing is superior to epicardial pacing with an even greater response achievable with optimization for each set of protocols. The mechanism of benefit may be due to the ability to access more optimal sites that cannot be reached by the constraints of the CS anatomy.
Furthermore, guidance to the optimal LV pacing site may be aided by modalities such as CMR to target nonscarred and delayed activating sites.