Author + information
- Received October 16, 2012
- Accepted November 15, 2012
- Published online August 1, 2013.
- Manav Sohal, MBBS∗,†,
- Anoop Shetty, MBChB∗,†,
- Simon Duckett, MD∗,
- Zhong Chen, MBBS∗,
- Eva Sammut, MBBS∗,
- Sana Amraoui, MD‡,
- Gerry Carr-White, MD, PhD†,
- Reza Razavi, MD∗∗ ( and )
- Christopher Aldo Rinaldi, MD∗,†
- ∗Division of Imaging Sciences and Biomedical Engineering, King's College, London, United Kingdom
- †Cardiovascular Directorate, Guy's and St. Thomas' NHS Foundation Trust, London, United Kingdom
- ‡Hôpital Cardiologique du Haut-Lévèque, Université Victor Segalen Bordeaux II, Bordeaux, France
- ↵∗Reprint requests and correspondence:
Prof. Reza Razavi, Division of Imaging Sciences, King's College London, Rayne Institute, Lambeth Wing, St. Thomas' Hospital, London SE1 7EH, United Kingdom.
Objectives Type II activation describes the U-shaped electrical activation of the left ventricle (LV) with a line of block in patients with left bundle branch block (LBBB). We sought to determine if a corresponding pattern of contraction could be identified using cardiac magnetic resonance (CMR) cine imaging and whether this predicted response to cardiac resynchronization therapy (CRT).
Background U-shaped LV electrical activation in LBBB has been shown to predict favorable response to CRT. It is not known if the degree of electromechanical coupling is such that the same is true for LV contraction patterns.
Methods A total of 52 patients (48% ischemic) scheduled for CRT implantation prospectively underwent pre-implantation CMR cine analysis using endocardial contour tracking software to generate time−volume curves and contraction propagation maps. These were analyzed to assess the contraction sequence of the LV. The effect of contraction pattern on CRT response in terms of reverse remodeling (RR) and clinical parameters (New York Heart Association functional class, 6-min walk distance and Heart Failure Questionnaire score) was assessed at 6 months.
Results Two types of contraction pattern were identified; homogenous spread from septum to lateral wall (type I, n = 27) and presence of block with a subsequent U-shaped contraction pattern (type II, n = 25). Rates of RR in those with a type 2 pattern were significantly greater at 6 months (80% vs. 26%, p < 0.001) as was mean increase in 6-min walk distance (126 ± 106 m vs. 55 ± 60 m; p = 0.004).
Conclusions Cine CMR can identify a U-shaped pattern of contraction which predicts increased echocardiographic and clinical response rates to CRT in patients with LBBB.
Cardiac resynchronization therapy (CRT) has been shown to improve quality of life, reduce heart failure hospitalizations, and reduce mortality in selected heart failure patients (1–3). It is indicated in patients with heart failure and electrical dyssynchrony, which usually manifests as left bundle branch block (LBBB) on surface electrocardiography (ECG) (3). QRS duration alone remains only a moderate predictor of response to CRT (4). Despite advances in patient assessment and implant technology, a significant proportion of patients do not respond (5) and noninvasive techniques for predicting CRT response, including echocardiographic assessment of dyssynchrony, have proved disappointing in multicenter studies (6).
The cornerstone of CRT is the correction of left ventricular (LV) dyssynchrony, but patients with LBBB may have marked heterogeneity of their LV activation pattern that may determine CRT response (7). Mapping of LV electrical activation in patients with LBBB has demonstrated 2 activation patterns: a U-shaped (type II) activation pattern with a line of conduction block between the septum and lateral wall, which is associated with a favorable response to CRT, and a more homogenous (type I) pattern associated with a lesser response to CRT (7–10). The demonstration of LV activation pattern generally requires methods such as contact and noncontact mapping (NCM), which are highly invasive and are associated with a significant risk and therefore are not widely applicable as useful clinical tools. Noninvasive methods that are able to reliably demonstrate the pattern of LV activation may therefore be clinically useful in predicting CRT response. Similarly, demonstration of a less favorable pattern of LV activation may alert the clinician to the need for nonstandard forms of CRT such as multisite or endocardial pacing.
Cardiac magnetic resonance (CMR) is able to accurately characterize myocardial contraction and motion. Assuming there is excitation-contraction coupling, CMR may be able to noninvasively evaluate LV activation patterns. CMR tagging and strain analyses give excellent spatial resolution when assessing cardiac motion, but tag decay remains a problem that makes it difficult to track motion through the entire cardiac cycle. We hypothesized that analysis of cine-CMR images using endocardial contour tracking software could identify LV contraction patterns analogous to LV activation. We investigated whether CMR derived LV contraction patterns could predict a favorable CRT response.
Study population and assessment
The study was approved by our institutional ethics committee, and all patients gave written informed consent. Prospective CMR assessment of 52 patients with conventional criteria for CRT (New York Heart Association [NYHA] functional classes II to IV, QRS duration >120 ms, and LV ejection fraction ≤35%) was performed. All patients underwent CMR imaging performed with a 1.5-T scanner (Achieva, Philips Healthcare, Best, the Netherlands) with a 32-element cardiac coil. Delayed-enhancement cardiac magnetic resonance was performed 15 to 20 min following the administration of 0.1 to 0.2 mmol/kg gadopentetate dimeglumine (Magnevist, Bayer Healthcare, Dublin, Ireland) using conventional inversion recovery techniques. Patients underwent standard transthoracic echocardiographic assessment using a Vivid 7 model scanner (General Electric-Vingmed, Milwaukee, Wisconsin). Analysis was performed with EchoPac version 6.0.1 (General Electric-Vingmed) providing data for LV function and volumes pre-implantation and at 6 months. Ejection fraction and LV dimensions were measured using the 2-dimensional (2D) modified biplane Simpson method. End-systolic volumes were recorded pre-implantation and at 6 months post-implantation. Pre-implant 3-dimensional (3D) echocardiographic assessment of dyssynchrony was performed to derive a 16-segment systolic dyssynchrony index (SDI). This was defined as the standard deviation of the time to reach minimum volume for each of the 16 segments of the LV expressed as a percentage of the cardiac cycle. Reverse remodeling (RR) was defined as a reduction in end-systolic volume (ESV) of ≥15%.
Assessment of the pattern of LV contraction/activation
A prototype software platform (TomTec, Unterschleissheim, Germany) was used to analyze long- and short-axis steady-state free precession CMR images. All CMR images were ECG gated, so all phases in all separate images were acquired at the same point in the cardiac cycle relative to the R-wave. The post-processing software mandates that the 4 sequences used (3 long-axis and 1 short-axis stack) have the same number of phases to ensure that each frame of each view reflects the same point in the cardiac cycle. The LV endocardial border was manually outlined at end-diastole and end-systole in 2-, 3-, and 4-chamber long-axis views. The workflow for the post-processing software takes an average of 10 to 12 min per patient. The software uses advanced algorithms to track endocardial motion producing a 3D shell of LV contraction. Time–volume curves were generated (Fig. 1) (11), and these were used to identify the latest activating segments on a modified 16-segment American Heart Association bulls-eye plot. A map depicting the propagation of contraction was generated based on the time–volume curves. A systolic dyssynchrony index (derived from the dispersion of time to minimum regional volume for all 16 LV segments) was also calculated (Fig. 1). Analogous to the pattern of LV electrical activation, the pattern of contraction was described as type I if propagation proceeded homogenously from the septum to lateral wall and type II if a line of block was apparent with a U-shaped pattern of propagation (Fig. 2). The CMR contraction patterns were assessed by 2 experienced independent observers blinded to the echocardiography results.
RR rates between patients with type I and type II patterns of contraction were compared. Additionally, changes in clinical parameters (NYHA functional class, 6-min walk distance, and Minnesota Living with Heart Failure Questionnaire score) were compared across the 2 groups.
Comparison with noncontact mapping
NCM data were available for 5 patients included in this analysis. NCM was performed prior to CRT implantation as part of a research protocol that has previously been published (10). A mapping balloon (Ensite Array EC1000 NCM, St. Jude Medical, St. Paul, Minnesota) was passed in a retrograde fashion across the aortic valve into the LV via the femoral artery. Analysis of isopotential maps and the associated morphological features of electrograms were used to determine the pattern of activation.
Statistical analysis was performed with PASW Statistics version 20 software (SPSS Inc., Chicago, Illinois). Group comparisons were performed using an independent-samples t test for normally distributed data after the data were tested for normality. Categorical variables were compared using a chi-square test. The Cohen kappa coefficient was used to test for interobserver variability of categorical variables. Interobserver agreements of continuous clinical measures were assessed according to the statistical methods proposed by Bland and Altman (12). The measure of reproducibility used was 2 SD of the interagreement indexes. Multivariate logistic regression analysis was used to evaluate independent predictors of CRT response. Values of p < 0.05 were considered statistically significant.
Fifty-two patients were studied with follow-up data available for the entire cohort. Table 1 shows the characteristics of the entire cohort and also grouped according to contraction pattern. All patients were in sinus rhythm. The majority of patients (88%) were male, with a relatively equal number of ischemic and nonischemic patients (48% vs. 52%, respectively). A CMR-derived type II contraction pattern was present in 25 of 52 (48%) patients. There was no significant difference between the groups according to age (64.7 ± 15.2 years vs. 64.9 ± 10.8 years, respectively) or sex. Nonischemic patients were significantly more likely to have a type II contraction pattern than ischemic patients (63% vs. 32%, respectively; p = 0.03). Most of the cohort was on excellent medical therapy, with no statistical differences in the number of patients taking angiotensin-converting enzyme inhibitors/angiotensin receptor blockers, β-blockers, or aldosterone antagonists when patients were divided according to contraction pattern. More patients with a type II contraction pattern were taking loop diuretics (92% vs. 67%, respectively; p = 0.02). Baseline NYHA functional class, 6-min walk distance, and Heart Failure Questionnaire scores were similar between the 2 groups. Patients with type II contraction tended to have broader QRS durations (type I 152 ± 20 ms vs. type II 159 ± 27 ms), but this did not reach statistical significance. SDI derived from 3D echocardiography was nonsignificantly greater in patients with type II contraction, whereas SDI derived from CMR was significantly greater in those with type II contraction (9.3 ± 3.9% vs. 13.2 ± 5.5%, respectively; p = 0.006).
Effect on RR
Table 2 shows the echocardiographic and clinical response rates for the cohort divided by contraction pattern. The response rate in terms of RR for the entire cohort was 52%. The mean change in echo-derived ESV at 6 months was significantly greater in the group with type II contraction (−21.3 ± 23.3% vs. −3.6 ± 25.7%, respectively; p = 0.02). The RR rate in those with a type II compared to type I pattern was significantly greater (80% vs. 26%, respectively; p < 0.001).
Effect on clinical response
The mean increase in 6-min walk distance was significantly greater in the group with a type II pattern (55 ± 60 m vs. 126 ± 106 m, respectively; p = 0.004). There was no statistically significant difference between the groups in terms of reduction in NYHA functional class or Heart Failure Questionnaire score.
Outcomes according to contraction pattern in ischemic and nonischemic patients
Further analysis according to contraction pattern was performed in patients with ischemic cardiomyopathy (ICM) and nonischemic cardiomyopathy (NICM) (Table 3). Patients were subgrouped according to cause and contraction pattern. In patients with NICM there was a significant increase in 6-min walk distance and a highly significant increase in RR in patients with a type II pattern (88% vs. 20%, respectively; p < 0.001). When compared to the entire cohort, NICM patients with a type II contraction pattern had a significantly improved response rate in terms of RR (Fig. 3). For ICM patients there was no significant difference in NYHA functional class reduction, Heart Failure Questionnaire score reduction, or increase in 6-min walk distance according to contraction pattern. RR rate was 29% versus 63% for type I and type II, respectively (p = 0.19).
Effect of scarring
Scar evaluation was possible in 47 of 52 patients (Table 1). Reduced renal function precluded the administration of gadolinium-based contrast in 4 patients, and 1 patient had a documented allergy to gadolinium. Scars of at least 25% transmurality in at least 2 segments were present in 26 (55%) subjects of the entire cohort in whom scar evaluation was possible. Of these 26 patients, 20 (77%) had at least 2 segments with full-thickness scar. Scars of at least 25% transmurality in at least 2 segments were present in 17 (71%) of those with a type I contraction pattern and 9 (31%) of those with a type II contraction pattern (chi-square = 4.78; p = 0.03). The same values for type I and II contraction when looking at transmural scar were 13 (54%) and 7 (30%), respectively (chi-square = 2.71; p = 0.10).
In a multivariate logistic regression model including QRS duration of >150 ms, cause, type II contraction pattern, and the presence of scars of at least 25% transmurality in at least 2 segments, a type II contraction pattern was the only statistically significant independent predictor of RR at 6 months, with an odds ratio of 8.39 (95% confidence interval: 1.86 to 37.8; p = 0.006) (Fig. 4).
Most LV pacing leads were positioned in either posterolateral or lateral branches of the coronary sinus (Table 1). There was no statistically significant difference in LV lead position when comparing those with a type I and type II pattern of contraction.
Correlation with noncontact mapping
CMR contraction maps were compared to activation maps derived from NCM in 5 patients. There was agreement between the 2 modalities in all 5 patients. Figure 5 shows an activation map derived from NCM in a patient with type II activation (Fig. 5, top panels) and the contraction front map derived from the analysis of the cine-CMR images. The anterior line of block and U-shaped pattern of activation identified by NCM is closely mirrored by the contraction front map.
Two observers assessed the CMR images and TomTec analyses and grouped patients according to the pattern of contraction. The kappa coefficient of agreement with regards to contraction pattern was 0.73, indicating excellent interobserver agreement. Further reproducibility analyses of endocardial contour segmentation were also performed. Two imaging experts independently segmented the endocardial contours of 10 datasets. The sequence of contraction was compared by assessing the time–volume curves (from which the contraction front maps are derived). The sequence was identical in all 10 datasets. There was a very close agreement in terms of CMR-derived SDI in each of the 10 datasets. The interobserver average difference was 0.4 ± 1.2%, and the coefficient of variance was 3.9 ± 3.0%.
We have shown that analysis of cine-CMR images with contour-tracking software can identify a pattern of LV contraction that is analogous to the pattern of LV activation described in patients with LBBB (8). Two patterns of contraction propagation were identified, one with a homogenous spread from the septum to lateral wall (type I) and the other with an apparent line of block, giving rise to a U-shaped pattern of propagation (type II). Patients with a type II pattern showed a more favorable response to CRT in terms of RR and increased 6-min walk distance. Subjective measures of CRT response were not significantly different, suggesting a possible placebo effect of CRT in the echocardiographic nonresponders.
Comparison with electrical activation derived from invasive mapping
In our patients, a type II pattern of contraction propagation was identified in 48% of patients. Auricchio et al. (8) was the first to describe LV activation patterns in patients with LBBB, demonstrating a type II pattern in 23 of 24 (96%) cases. Subsequent studies have found the prevalence of type II activation to be lower (7,10,13). Lambiase et al. (7) found a type II activation pattern in 6 of 10 (60%) patients studied (13), and Fung et al. (7) identified a type II activation pattern in 15 of 23 patients (65%). Our relatively low percentage of a type II contraction pattern may be explained by several factors. First, compared to previous studies, ours had a greater proportion of ischemic patients, a group in which type I activation is more prevalent (7). Second, it is possible that the CMR-derived contraction pattern may not have the spatiotemporal resolution to correctly identify a type II activation pattern in our patients (however, our NCM data, albeit in a small number of patients, would suggest a good correlation). Our echocardiographic RR rate of 52% is lower than that of Fung et al. (7) at 61% and this may reflect the higher proportion of ischemic patients in the current study. Of note, the rate of RR in the study by Fung et al. (7) was 80% for type II and 25% for type I, which is strikingly similar to our findings of 80% and 26%, respectively.
Mechanical contraction and electrical activation
Our study used mechanical contraction pattern as a surrogate for electrical activation. The question remains as to whether it is reasonable to assume an electromechanical association such that the type II electrical activation pattern is reflected in a similar pattern of mechanical propagation. Animal studies have shown a discrepancy between electrical and mechanical activation (14,15), with mechanical delay exceeding electrical delay under conditions of ventricular pacing, a finding later confirmed in a canine model of LBBB (16). Another canine study showed that prolonged delay from regional electrical activation to the onset of shortening in late-activated segments during LBBB is the result of an increased rate in pressure rise (LV dP/dt) at the time of activation and that LV contractility is positively correlated with increasing electromechanical delay (17). It appears, therefore, that mechanical delay may overestimate the extent of electrical dyssynchrony, but there is no evidence to suggest a difference in the pattern of electrical and mechanical activation. Indeed, a linear relationship between electrical activation and contraction onset in healthy paced canine hearts has been demonstrated (18).
Fung et al. (7) were the first to provide information on an electromechanical association in humans by combining echocardiographic and endocardial mapping data in the same patients receiving CRT. They found a close correlation between the LV activation time and echocardiographic markers of mechanical dyssynchrony in patients with a type II activation but not in patients with a type I pattern (7). In keeping with this, our patients with type II contraction had a significantly greater degree of CMR-derived dyssynchrony (SDI) than those with a type I contraction pattern. The echo-derived SDI difference between contraction patterns also had a tendency toward more mechanical dyssynchrony in those with a type II pattern, but this did not reach statistical significance. This difference between CMR- and echo-derived SDI is likely explained by 2 factors. First, endocardial definition is superior with CMR, thereby allowing better identification of the contours. Second, the echo data calculate the SDI using a full 3D image of the LV, whereas CMR extrapolates data from combining multiple 2D short-axis and long-axis views. A lesser degree of electromechanical coupling in patients with type I activation may explain the lower response rates to CRT in this group. Our study provides some degree of electrical and mechanical correlation, albeit in a small cohort of patients. Five patients in our series had NCM data available, and the isopotential maps of electrical activation matched the contraction maps (Fig. 5). Our data suggest an electromechanical association and that the presence of a type II pattern of mechanical contraction propagation predicts a favorable response to CRT, particularly in NICM. A notable finding of the present study was the poor response rate to CRT in NICM with type I activation (20% RR), suggesting that identification of this contraction pattern may be especially important in nonischemic patients. The presence of scar might be expected to negate the ability of CMR to identify a type II pattern of contraction. A type II pattern was identified in 32% of ischemic patients, and there was a tendency toward increased RR rates in this group. Statistical significance was not reached and may be because of the small absolute number in that group. If this finding is verified in larger studies it may be useful as a means of identifying which patients are more likely to respond in what is an extremely challenging group of patients to treat.
Other noninvasive modalities for assessing LV activation
Electrocardiographic imaging has emerged as a potential technique for noninvasively assessing LV activation. It is based on 250 body surface ECGs and a patient-specific heart-torso anatomy derived from an ECG-gated computed tomographic scan. Electroanatomic maps of epicardial potentials, electrograms, and activation sequences are generated. In a study of 25 patients with NICM and LBBB, responders to CRT were found to have a line of conduction block located between the epicardial aspect of the septum and LV lateral wall (19). Whether this finding extends to patients with ICM is not yet known. Our study (and the CRT population as a whole) includes a large proportion of patients with ICM, and our findings suggest that if a patient has an ischemic cause and a type I pattern of LV contraction propagation, then the patient is unlikely to respond to conventional CRT. An alternative ultrasound-based modality, electromechanical wave imaging, has shown some potential in mapping electromechanics of all 4 cardiac chambers at high temporal and spatial resolution in both animals and humans (20). A linear relationship between electromechanical wave and electrical activation sequences has been shown, thereby demonstrating that the electromechanical wave follows the electrical activation sequence in healthy tissue. Whether these findings will be replicated in diseased hearts remains to be seen.
We have used noninvasive assessment of the contraction pattern from CMR as a surrogate for electrical activation. The accuracy of this technique is a potential limitation. The software used in this study generates time–volume curves and contraction propagation maps by contouring endocardial motion on cine-CMR images. It is recognized that traditional cardiac imaging pulse sequences such as tagged CMR and phase-contrast CMR contain more information about regional myocardial motion than cine-CMR. Temporal resolution of tagging sequences currently is on the order of 15 to 20 ms, which is sufficient to detect peak systolic strain, and this will potentially miss movement in the isovolumic contraction period, which is often seen in LBBB (21). NCM data were available only for a small proportion of the cohort, but a growing body of evidence suggests that the sequence of mechanical activation follows that of electrical activation, and the demonstration of a good association between the MR-derived contraction front map and the electrical activation maps obtained from NCM in some of our cohort suggests that analysis of cine-CMR images holds promise as a means for detecting LV contraction patterns that predict enhanced CRT response rates.
The number of patients in our analysis was relatively small (although larger than any other study looking at activation patterns in CRT patients), and an appropriately powered randomized trial is necessary to ensure that the findings in this study are more generally applicable.
Analysis of cine-CMR sequences using dedicated software to generate time–volume curves and contraction front maps can demonstrate 2 types of LV mechanical activation which mirror that reported for electrical activation. Mechanical activation characterized by a line of block between the septum and LV lateral wall predicted an increased response rate to CRT, particularly in nonischemic patients. Noninvasive identification of contraction patterns using cine-CMR may prove a clinically useful tool to identify a group of patients that will benefit from CRT.
Drs. Sohal and Shetty have received an educational grant from St. Jude Medical. Dr. Razavi has received investigator-led grant funding from Philips Healthcare. Dr. Rinaldi is a consultant to St. Jude and has received research funding from St. Jude, Medtronic, and Boston Scientific. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- cardiac magnetic resonance
- cardiac resynchronization therapy
- ischemic cardiomyopathy
- left bundle branch block
- noncontact mapping
- nonischemic cardiomyopathy
- Received October 16, 2012.
- Accepted November 15, 2012.
- American College of Cardiology Foundation
- Ypenburg C.,
- van Bommel R.J.,
- Borleffs C.J.,
- et al.
- Chung E.S.,
- Leon A.R.,
- Tavazzi L.,
- et al.
- Fung J.W.,
- Chan J.Y.,
- Yip G.W.,
- et al.
- Auricchio A.,
- Fantoni C.,
- Regoli F.,
- et al.
- Duckett S.G.,
- Camara O.,
- Ginks M.R.,
- et al.
- Lambiase P.D.,
- Rinaldi A.,
- Hauck J.,
- et al.
- Wyman B.T.,
- Hunter W.C.,
- Prinzen F.W.,
- McVeigh E.R.
- Russell K.,
- Opdahl A.,
- Remme E.W.,
- et al.
- Russell K.,
- Smiseth O.A.,
- Gjesdal O.,
- et al.
- Badke F.R.,
- Boinay P.,
- Covell J.W.
- Provost J.,
- Lee W.N.,
- Fujikura K.,
- Konofagou E.E.