Outcomes of Cardiac Resynchronization Therapy with Image-Guided Left Ventricular Lead Placement at the Site of Latest Mechanical Activation: A Systematic Review and Meta-Analysis

Aim To assess evidence for an image-guided approach for cardiac resynchronization therapy (CRT) that targets left ventricular (LV) lead placement at the segment of latest mechanical activation. Methods A systematic review of EMBASE and PubMed was performed for randomized controlled trials (RCTs) and prospective observational studies from October 2008 through October 2020 that compared an image-guided CRT approach with a non-image-guided approach for LV lead placement. Meta-analyses were performed to assess the association between the image-guided approach and NYHA class improvement or changes in end-systolic volume (LVESV), end-diastolic volume (LVEDV), and ejection fraction (LVEF). Results From 5897 citations, 5 RCTs including 818 patients (426 image-guided and 392 non-image-guided) were identified. The mean age ranged from 66 to 71 years, 76% were male, and 53% had ischemic cardiomyopathy. Speckle tracking echocardiography was the primary image-guided method in all studies. LV lead placement within the segment of the latest mechanical activation (concordant) was achieved in the image-guided arm in 45% of the evaluable patients. There was a statistically significant improvement in the NYHA class at 6 months (odds ratio 1.66; 95% confidence interval (CI) [1.02, 2.69]) with the image-guided approach, but no statistically significant change in LVESV (MD −7.1%; 95% CI [−16.0, 1.8]), LVEDV (MD −5.2%; 95% CI [−15.8, 5.4]), or LVEF (MD 0.68; 95% CI [−4.36, 5.73]) versus the non-image-guided approach. Conclusion The image-guided CRT approach was associated with improvement in the NYHA class but not echocardiographic measures, possibly due to the small sample size and a low rate of concordant LV lead placement despite using the image-guided approach. Therefore, our meta-analysis was not able to identify consistent improvement in CRT outcomes with an image-guided approach.


Introduction
CRT has been shown to improve outcomes in heart failure patients; however, even among the carefully selected patients for whom clinical trials and subsequent guidelines support the use of CRT, not all patients realize the full benefits of this therapy [1][2][3]. e left ventricular (LV) lead placement has been identified as an important factor for CRT response [4][5][6][7]. In addition to avoiding placement in an apical segment, placing the LV lead in the LV segment with the latest mechanical activation that is free from transmural scarring has been suggested [5,8,9]. Previous studies using speckle tracking echocardiography (STE) to identify and target the LV segment of latest mechanical activation showed promising results but were relatively small studies, including 1 to 2 centers with unclear generalizability [10,11].
us, the question remains as to whether a personalized approach to LV lead placement that uses cardiac imaging to identify and target the LV site of the latest mechanical activation can improve CRT outcomes. To address this question, we conducted a systematic literature review to identify randomized controlled studies or prospective observational studies of image-guided approaches for LV lead placement and performed meta-analyses to assess the association between the image-guided approach and CRT outcomes.

Methods
is review was conducted as part of a National Heart, Lung, and Blood Institute-funded project to synthesize evidence related to CRT and to identify and prioritize clinical and policy evidence gaps [12]. To address one component of the evidence gap regarding the extent and/or location of LV mechanical dyssynchrony predicting CRT outcomes, we conducted a systematic literature review for studies evaluating CRT outcomes in patients with an image-guided approach for LV lead placement within the LV segment with the latest mechanical activation versus a non-image-guided approach. e Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement was followed [13].

Literature Search Strategy.
A literature search was conducted using PubMed and Embase for randomized controlled trials (RCT) or prospective studies of CRT. e MeSH and Emtree terms "cardiac resynchronization therapy" and "cardiac resynchronization therapy device" were used in addition to title and abstract searches for "cardiac resynchronization therapy," "cardiac resynchronization therapy," "atrio biventricular pacing," "atrio-biventricular pacing," "biventricular pacing," or "biventricular pacemaker." Results were limited to English language articles and human studies with publication dates between October 2008 and October 2020.

Study Selection, Abstraction, and Bias Assessment.
Two reviewers independently screened all identified titles and abstracts. Publications that met the selection criteria as determined by either reviewer were moved on to full-text review. In addition to being either an RCTor prospective observational study in humans and published in English, all studies had to include the following: the use of CRT in eligible inpatient or outpatient heart failure patients; an assessment of the location and/or extent of LV mechanical dyssynchrony by any imaging method; a report of CRT outcomes of interest in relation to the LV mechanical dyssynchrony. CRT outcomes of interest included all-cause mortality, heart failure mortality, heart failure hospitalization, ventricular arrhythmias, change in LV dyssynchrony, change in echocardiographic parameters (LV ejection fraction, end-systolic volume or diameter, end-diastolic volume or diameter), NYHA class improvement, 6minute walk test (6 MWT), ICD shocks, quality of life, devicerelated adverse events, and any composite of the outcomes. Two reviewers then independently reviewed the full text of each selected publication to confirm that the study met the inclusion criteria. Finally, one investigator reviewed the full text of the selected studies to identify those that compared CRT outcomes in patients with an image-guided approach for LV lead placement at the site of the latest LV mechanical activation to those without an image-guided approach for LV lead placement.
One investigator abstracted study characteristics, patient characteristics, and results, and a second investigator confirmed study applicability and the accuracy of the abstracted data. Abstracted study characteristics included study design, number of study sites, location of study site (s), funding source, publication year, study arms, number of subjects, method of assessing LV mechanical dyssynchrony, imaging method, follow-up period, and outcomes. Patient characteristics included age, sex, ischemic cardiomyopathy versus non-ischemic cardiomyopathy, LBBB, QRS duration, sinus rhythm, LV ejection fraction, NYHA class, and location of LV lead. In addition, if reported, the number of patients whose LV lead was confirmed to have been placed within the LV segment of latest mechanical activation (concordant LV lead placement) or next to the segment of latest mechanical activation (adjacent LV lead placement) in each study arm was also captured. e abstracting investigator and the over-reading investigator independently assessed risk of bias using the revised Cochrane risk-of-bias tool for randomized trials [14]. Discrepancies were resolved through discussion.

Statistical Analysis.
Categorical outcomes (≥1 class improvement in the NYHA class) were reported as frequencies with percentages. Continuous values were reported as means with standard deviations (SDs). For LVEF, the mean (SD) of the absolute change from baseline to follow-up was reported. For LVEDV and LVESV, the mean (SD) of the relative change, expressed as a percentage, from baseline was reported. For one study, the mean relative change and standard deviation in LVEDV and LVESV were estimated using the reported mean and standard deviation at baseline and the absolute change from baseline since the relative change was not reported [10].
Meta-analyses were conducted for all outcomes of interest that were reported in three or more included studies. Meta-analyses were performed using a DerSimonian-Laird random-effects model, and we conservatively used the Knapp-Hartung approach to adjust the standard errors of the estimated model coefficients [15,16]. Categorical outcomes were pooled and presented as an odds ratio (OR) with 95% confidence intervals (CIs). Continuous outcomes were pooled and presented as the mean difference (MD) with 95% CI. Heterogeneity was assessed with Cochrane's Q and I 2 indexes and the corresponding p value. I 2 values greater than 75% indicate large heterogeneity.

Results
From 3084 unique citations identified, a total of 5 studies met the selection criteria for this analysis ( Figure 1) [10,11,[17][18][19]. All studies were randomized controlled trials, and all but one were single-center studies [10]. While the selected imaging methodology in each study to identify the site of latest mechanical activation was performed on all patients enrolled in the study, the results were provided to the implanting physician only for those patients randomized to the imaging arm. Speckle tracking echocardiogram (STE) was the primary imaging method used in all studies to identify the LV segment of latest mechanical activation and thus the targeted LV segment for LV lead placement. However, three of the studies use multimodal imaging to further refine the targeted segment [17][18][19]. All three used computed tomography (CT) to visualize coronary sinus branches. In addition, one study used cardiac magnetic resonance (CMR) [17], one used rubidium positron emission tomography [18], and one used single-photon emission computed tomography [19] to identify areas of transmural myocardial scar to avoid LV lead placement. In the control arms, CRT placement was as per the standard of care in three studies [10,11,17] and was electrically guided in two studies [18,19]. However, in one study with the standard-of-care CRT placement, results from the CT and CMR were also made available to physicians [17]. Four of the studies were rated as having a low overall risk of bias [10,[17][18][19], and one was rated as having some overall risk of bias [11]. Study characteristics are summarized in Table 1. A total of 818 patients were enrolled in the 5 identified RCTs, with 426 (52%) in the intervention arm and 392 (48%) in the comparator arm. Patient characteristics are presented in Table 2. Characteristics were similar between the control and intervention arms of each study. e mean patient age was greater than 65 years in all studies, and most patients were male (range of 73% to 79%). e proportion of patients with ischemic cardiomyopathy in each study ranged from 46% to 62%, the mean or median QRS duration was over 150 msec in all studies, and all but one study enrolled patients with NYHA classes II, III, and IV. Regardless of the study arm, most LV leads were ultimately placed within a lateral or posterior segment (Table 2). Among 390 evaluable patients in the image-guided arm, 176 (45%) had concordant LV lead placement and 172 (44%) had the LV lead placed in an adjacent segment. In the non-image-guided arm, 117 of 366 (32%) patients had a concordant lead placement and 174 (48%) had an adjacent lead placement.
All outcomes of interest reported in each study are listed in Table 1; however, only four outcomes of interest were  of outcome, F � selection of reported results, and G � overall (green indicates "low", yellow indicates "some", and red indicates "high" risk of bias).

Discussion
In our meta-analysis, we found a statistically significant improvement in the NYHA class with the image-guided approach as compared to the non-image-guided approach; however, no statistically significant differences in any of the echocardiographic measures were found between the approaches. Mortality and heart failure hospitalization outcomes were reported in four of the five included studies; however, because the reported outcome measures were inconsistent across studies and/or the number of events in each study arm could not be ascertained from the publication, meta-analyses for these outcomes were not feasible. e results of our meta-analysis are not consistent with a prior meta-analysis; however, as described in the following, our study provides an expanded and more contemporary perspective [20]. In addition, the five RCTs included in our study had differences in study enrollment criteria and components of the intervention and/or comparator arms, leading to some heterogeneity; however, these trials mirror the evolution of technology and guideline recommendations for CRT. Lastly, the proportion of patients in each study who achieved LV lead placement in the targeted segment varied, as did the rate of patients in the comparator arms who fortuitously had the LV lead placed in the LV segment of latest activation without a scar. is may have resulted in a diminished added value with the imaging-guided approach.   erefore, our meta-analysis was not able to identify consistent improvement in CRT outcomes with an imageguided approach. e systematic review and meta-analysis conducted by Jin et al. included two studies using speckle tracking echocardiography to identify the target for optimal LV lead placement (Targeted Left Ventricular Lead Placement to Guide CRT (TARGET) and Speckle Tracking Assisted Resynchronization erapy for Electrode Region (STARTER) studies) and one study using intracardiac echocardiography (ICE) during LV lead implantation to select the best placement option [10,11,20,21]. Jin et al. concluded that the image-guided approach was associated with a statistically significant greater CRT response (OR 2.10; 95% CI 1.43 to 3.07), improvement in LVEF (MD 3.46; 95% CI 1.91 to 5.01), and reduction of LVESV (MD −20.36; 95% CI −27.82 to −12.90) at 6 months [10,11,20,21]. CRT response was defined differently in the three included studies, and therefore, this endpoint was not included in our analysis. In addition, we excluded the study using ICE because it did not meet our inclusion criteria for an imageguided approach that identified a specific target for LV lead placement [21]. However, our analysis included three additional studies that were published after the meta-analysis by Jin et al. [17][18][19]. ese newer studies included multimodal imaging approaches in the intervention arm, and two of them included an electrically guided approach as the comparator. e results from these three studies, in contrast to TARGET and STARTER, largely showed no difference in CRT outcomes between the image-guided and non-imaged  e five RCTs in our study were published over an approximately 8-year time span, and thus, patient selection criteria and components of the image-guided or comparator arms evolved with technology and guidelines during this time. TARGET and STARTER were the first RCTs to address whether an image-guided approach for LV lead placement can improve CRT outcomes [10,11]. Patients with a QRS duration of ≥120 milliseconds (ms) without considering the presence of LBBB were enrolled. In fact, neither study presents the number of patients with LBBB. However, subsequent studies, following the evolution of clinical practice guidelines, enrolled a high proportion of patients having LBBB, for whom a standard approach targeting the posterior or posterolateral LV segment may be more likely to be the segment of latest mechanical activation [17][18][19]. In addition, TARGET and STARTER used STE as the sole imaging method. Subsequent studies incorporated multimodal imaging into the intervention arm to better identify transmural scar and coronary anatomy [17][18][19]. Likewise, the components of the comparator arm evolved over time to include electrically guided approaches and/or some imaging results to inform on coronary anatomy [17,18]. is evolution in methodology may have resulted in greater heterogeneity among studies and diminishment of the valueadded response with an image-guided approach. However, these newer studies more closely represent contemporary capabilities and the importance of this issue, as further reinforced by the recent initiation of a small RCT in the Netherlands comparing CRT outcomes in those with realtime image-guided LV lead placement using CMR feature tracking versus those with the standard-of-care LV lead placement with electrical guidance [22].
Lastly, the primary goal of an image-guided approach is to place the LV lead within the LV segment with latest mechanical activation that is free of scar. While electrical guidance and programming are important components for CRT outcomes, those alone are not likely able to overcome issues associated with a less-than-optimally placed LV lead [22,23]. Imaging to identify the site of latest mechanical activation was conducted in all patients in all of the included studies; however, the imaging data were only provided to the implanting physician for those patients in each study who were randomized to the image-guided study arm. erefore, post hoc, the number of patients who ultimately had their LV lead placed at the site of latest mechanical activation (concordant), in the segment adjacent to the segment with the latest mechanical activation (adjacent), and distant from the segment of the latest mechanical activation (discordant) in both study groups was ascertainable and was reported in all of the included studies. e proportion of patients who achieved a concordant LV lead varied among studies with the highest proportion at 61% in TARGET and the lowest proportion at 21% in the study by Borgquist et al. In addition, the proportion of patients with the fortuitous placement of the LV lead within the segment with the latest mechanical activation, free of scar, in the comparator arms also varied, with the highest proportion at 45% in TARGET and the lowest proportion at 12% in STARTER. In a separate meta-analysis, we also found a statistically significant association between CRT outcomes and concordant or adjacent LV lead placement versus discordant LV lead placement [24]. erefore, the potential lack of differentiation in concordant LV lead placement between study arms may have also diminished the effect from the intervention and merit further study.
ere are several limitations to this study, including the relatively small number of studies eligible for inclusion and the small number of total patients. All but one study was a single-center study; however, the included multicenter study only included 2 sites. erefore, performance of the intervention and the results may not be generalizable. ere was    also considerable heterogeneity identified in the analyses for LVEF and LVESV. As described above, changes in guideline recommendations and technology over time resulted in potentially significant differences in study populations and/ or LV lead implantation techniques, resulting in heterogeneity across studies and/or within studies. Lastly, the study by Borgquest et al. was terminated early due to equivocal results between study arms, potentially introducing bias [17].
In conclusion, our meta-analysis found that an imageguided CRT approach was associated with improvement in the NYHA class but not echocardiographic measures. erefore, our meta-analysis was not able to identify consistent improvement in CRT outcomes with an image-guided approach. While the small sample size and potential lack of differentiation between study arms in the achievement of concordant LV lead placement may partially explain the equivocal findings, it is also important to note that optimal LV lead placement alone may not fully address the complexity of heart failure management with CRT. Other factors such as device optimization, percentage of biventricular pacing, and arrhythmia burden may also need to be considered and integrated into any future strategic approach for CRT.

Data Availability
e data underlying the findings are included in the article.

Additional Points
Summary and Highlights. Five randomized controlled trials were identified that compared CRToutcomes with an imageguided approach versus a non-image-guided approach; however, the components of the image-guided approach and non-image-guided approach were not identical in all studies. Achievement of a left ventricular (LV) lead placement within the LV segment with the latest mechanical activation (concordant) occurred in 45% of evaluated patients with the image-guided approach but also fortuitously occurred in 32% of evaluated patients with the non-image-guided approach. e image-guided approach for placement of the LV lead for CRT within the LV segment with the latest mechanical activation was associated with a statistically significant greater odds of the NYHA class improvement at 6 months than the use of a non-image-guided approach. No differences in echocardiographic measures at 6 months following the CRT implant were found with the imageguided approach as compared to a non-image-guided approach. erefore, this meta-analysis was not able to identify consistent improvement in CRT outcomes with an imageguided approach.

Disclosure
Dr. Al-Khatib receives research funding from Medtronic, Abbott, and Boston Scientific.