Between 2 and 6 % of LV leads cannot be placed via the transvenous approach and require epicardial lead placement via thoracotomy [34]. Between 4 and 10 % of patients experience a clinically significant lead dislodgement necessitating repositioning [17]. The incidence of other complications has been assessed and reported in a recent review in the Journal of the American Medical Association [17]. In 54 reviewed studies including over 6000 patients the implant success rate was 93 % with peri-procedural complications occurring in 4 % of cases and peri-procedural death occurring in 0.3 % of patients [17]. Coronary sinus dissection or perforation can occur, but is rarely a fatal complication owing to the relatively low pressure of the cardiac venous system with pericardial effusion/tamponade complicating only 0.6–1.2 % of cases. Just over 1 % of patients developed device site infections consistent with the reported risk of infectious complications of 1–2 % for all implanted electrophysiologic devices.

One of the challenges in assessing CRT efficacy in clinical trials has been disagreement regarding what outcomes should be used to assess response. Fornwalt et al. recently assessed the agreement among the 15 most common response-criteria used in 26 published CRT studies. They found that 99 % of patients in these studies were CRT “responders” on the basis of at least 1 criteria, but that 75 % of comparisons between response-criteria demonstrated poor agreement [35]. The selection of response-criteria also has important implications on trial design as endpoints such as quality of life, 6-min walk, and peak VO₂ are either subjective or effort dependent measures that are susceptible to bias. Substudy analysis from MADIT-CRT began to identify characteristics associated with patients who are “super-responders” to CRT [36]. Table 17.2 presents other clinical features demonstrating efficacy in predicting response to CRT along with supporting references [36–44]. Importantly, to date, no features have demonstrated more reliability in predicting response to CRT than QRS duration and the presence of LBBB [45].

While the majority of appropriately selected patients derive clinical benefit from CRT, approximately 30 % of patients with appropriate QRS criteria are nonresponders without improvement in symptoms, ventricular remodeling, or HF hospitalizations. Mullens et al. demonstrated that a multi-disciplinary CRT nonresponder clinic could successfully identify and intervene on omissions or insufficiencies in the care of CRT nonresponders in a positive way. Interventions addressed suboptimal AV delay, loss of BiV pacing due to arrhythmia, suboptimal LV lead position, and suboptimal HF pharmacotherapy [46].

In the case of patients who do respond to CRT it is critical that the augmentation of blood pressure and cardiac performance provided by CRT be used as an opportunity to titrate neurohormonal antagonists. Small restrospective studies have demonstrated that CRT facilitates titration of BB and ACEI with simultaneous weaning of diuretics [47].

Atrial fibrillation (AF) is exceedingly common in patients with HF and becomes more prevalent with increasing degrees of HF severity. The OPTIME CHF and CONSENSUS studies assessing the utility of milrinone and enalapril respectively in patients with NYHA IV symptoms demonstrated a prevalence of 35–50 % [48]. It is therefore not surprising that AF is amongst the most common reasons for loss of BiV pacing by facilitating rapid AV conduction that outpaces programmed AV delays or exceeding the upper tracking limit. For this reason CRT trials have uniformly excluded patients with AF.

Particularly in patients with AF it is critical to confirm biventricular pacing by ECG as device diagnostics may spuriously report BiV pacing. Kamath et al. demonstrated that 12-lead Holter monitoring in patients with permanent atrial fibrillation frequently revealed inadequate BiV pacing percentages in patients with device diagnostics reporting very high rates of BiV pacing [49]. Approaches to AF in CRT include algorithms to achieve higher BiV pacing rates including ventricular sensed response, conducted AF response, and atrial tracking recovery; intensification of rate control; anti-arrhythmic drug therapy; and ultimately AV nodal ablation. Ganesan et al. performed a meta-analysis examining the utility of AVN ablation in AF patients with CRT. They demonstrated that AVN ablation was associated with significant reductions in all-cause and cardiovascular mortality, as well as improvement in mean NYHA functional class when compared to rate control strategies [50]. The updated ACC/AHA/HRS guidelines provide a IIa indication for CRT in these patients only if appropriate measures are taken to achieve high rates of BiV pacing (Table 17.1).

Currently, only a minority of HF patients meet the QRS criteria for CRT and not all CRT candidates are clinical responders. As such, there are ongoing efforts to identify measures of mechanical dyssynchrony that can assist in better predicting response. Beshai et al. assessed the utility of CRT in 172 patients with narrow QRS complexes but with echocardiographic evidence of mechanical dyssynchrony (defined as opposing-wall delay of ≥65 ms on tissue Doppler imaging or a mechanical dyssynchrony in the septal-to-posterior wall of ≥130 ms) in the ReithinQ trial. They found that while CRT treated patients had a significant improvement in NYHA class there was no significant improvement in quality of life, 6-min walk, LV reverse remodeling, or the primary endpoint of peak VO₂ [51]. The Predictors of Response to CRT (PROSPECT) trial assessed the efficacy of 12 echocardiographic parameters of mechanical dyssynchrony in predicting CRT response in 498 patients as evinced by an improvement in clinical composite score and a ≥15 % reduction in the LV end systolic volume at 6 months. In general these parameters demonstrated poor sensitivity and specificity with receiver operating characteristic area under the curves (ROC AUC) for most parameters falling between 0.5 and 0.6 [52].

In contrast to these disappointing results, the Speckle Tracking and Resynchronization (STAR) study assessed the ability of radial, circumferential, transverse, and longitudinal strain evidence of mechanical dyssynchrony (≥130 ms opposing wall delay) to predict improvements in LVEF and adverse long-term events including death, transplant, or left ventricular assist device therapy. Radial and transverse dyssynchrony predicted improved LVEF response. Interestingly, patients undergoing CRT without transverse or radial dyssynchrony had a significantly higher rate of adverse HF endpoints [37]. In the NARROW-CRT study, Muto et al., demonstrated that in patients with ischemic cardiomyopathy, narrow QRS complexes and echocardiographic evidence of mechanical dyssynchrony (≥60 ms difference in septal and lateral time-to-peak systolic velocity), CRT-D resulted in improvements in clinical composite scores and a reduction in the composite endpoint of HF hospitalization, HF death, and spontaneous ventricular fibrillation [41]. While intriguing STAR and NARROW-CRT were small studies and require corroboration in larger scale trials. While data has not yet been published, the largest study to date investigating CRT in patients with narrow QRS complexes, EchoCRT was terminated early due to futility after recruiting over 1000 patients [53].