Echocardiographic Quantification of Left Ventricular Asynchrony Predicts an Acute Hemodynamic Benefit of Cardiac Resynchronization Therapy

Journal of the American College of Cardiology Vol. 40, No. 3, 2002 2002 by the American College of Cardiology Foundation ISSN 0735-1097/02/$22.00 Published by Elsevier Science Inc. PII S0735-1097(02)01987-3 Echocardiographic Quantification of Left Ventricular Asynchrony Predicts an Acute Hemodynamic Benefit of Cardiac Resynchronization Therapy Pacing and Heart Failure Ole A. Breithardt, MD,* Christoph Stellbrink, MD,* Andrew P. Kramer, PHD, Anil M. Sinha, MD,* Andreas Franke, MD,* Rodney Salo, MSC, Bernhard Schiffgens, BSC,* Etienne Huvelle, MD, Angelo Auricchio, MD, PHD, for the PATH-CHF Study Group Aachen and Magdeburg, Germany; St. Paul, Minnesota; and Brussels, Belgium OBJECTIVES BACKGROUND METHODS RESULTS CONCLUSIONS We sought to determine whether radial left ventricular (LV) asynchrony in patients with heart failure predicts systolic function improvement with cardiac resynchronization therapy (CRT). We quantified LV wall motion by echocardiography to correlate the effects of CRT on LV systolic function with wall motion synchrony. Thirty-four patients underwent echocardiographic phase analysis of LV septal and lateral wall motion and hemodynamic testing before CRT. Phase relationships were measured by the difference between the lateral ( L ) and septal ( S ) wall motion phase angles: LS L S. The absolute value of LS was used as an order-independent measure of synchrony: LS L S. Three phase relationships were identified (mean SD): type 1 (n 4; peak positive LV pressure [dp/dt max ] 692 310 mm Hg/s; LS 5 6, synchronous wall motion); type 2 (n 17; dp/dt max 532 148 mm Hg/s; LS 77 33, delayed lateral wall motion); and type 3 (n 13; dp/dt max 558 154 mm Hg/s; LS 115 33, delayed septal wall motion, triphasic). A large LS predicted a larger increase in dp/dt max with CRT (r 0.74, p 0.001). Sixteen patients were studied during right ventricular (RV), LV and biventricular (BV) pacing. Cardiac resynchronization therapy acutely reduced LS from 104 41 (OFF) to 86 45 (RV; p 0.14 vs. OFF), 71 50 (LV; p 0.001 vs. OFF) and 66 42 (BV; p 0.001 vs. OFF). A reduction in LS predicted an improvement in dp/dt max in type 2 patients for LV (r 0.87, p 0.005) and BV CRT (r 0.73, p 0.04). Echocardiographic quantification of LV asynchrony identifies patients likely to have improved systolic function with CRT. Improved synchrony is directly related to improved hemodynamic systolic function in type 2 patients. (J Am Coll Cardiol 2002;40:536 45) 2002 by the American College of Cardiology Foundation QRS prolongation in left bundle branch block (LBBB) is associated with asynchronous ventricular contraction and a depressed ejection fraction (1) and is inversely correlated to global contractile function (2). Cardiac resynchronization therapy (CRT) has been recently introduced as a complementary treatment for patients with heart failure and a ventricular conduction delay, and it has been shown to improve left ventricular (LV) systolic function, as measured by peak positive LV pressure (dp/dt max ) (3) and Doppler echocardiography (4). It improves clinical symptoms and may lead to reversal of LV remodeling (5). From the *Department of Cardiology, University Hospital, Aachen, Germany; Guidant Corporation, St. Paul, Minnesota; Guidant Corporation, Brussels, Belgium; and Department of Cardiology, University Hospital, Magdeburg, Germany. This work was supported by a grant from Guidant Corporation, Brussels, Belgium. The investigators and participating centers of the Pacing Therapies for Congestive Heart Failure (PATH-CHF) Study Group, along with collaborators from the Guidant CHF Research Group, are listed in the Appendix of Circulation 1999;99: 2993 3001. Manuscript received February 11, 2002; revised manuscript received April 1, 2002, accepted April 4, 2002. It is assumed that CRT improves systolic ventricular function by restoring more synchronized contraction patterns. However, only a few studies have investigated CRT mechanisms by using direct measures of ventricular asynchrony. Studies using multiple-gated equilibrium bloodpool scintigraphy demonstrated reduced interventricular phase shifts between the LV and right ventricular (RV) contraction sequence with CRT, but had conflicting conclusions about whether CRT reduced intraventricular asynchrony (6,7). Tagged magnetic resonance imaging has been used to quantify baseline mechanical dyssynchrony (8), but this modality is not applicable to patients with a pacemaker. In contrast, echocardiography allows rapid bedside evaluation of cardiac function and ventricular wall motion abnormalities. Abnormal septal wall motion patterns in patients with LBBB undergoing ventricular pacing have been studied by M-mode echocardiography, but these measurements are limited to the evaluation of radial function in the basal LV segments using the parasternal views. Recently, im-

JACC Vol. 40, No. 3, 2002 August 7, 2002:536 45 Breithardt et al. Quantification of Ventricular Asynchrony in CRT 537 Abbreviations and Acronyms ANOVA analysis of variance AV atrioventricular BV biventricular CAD coronary artery disease CRT cardiac resynchronization therapy DCM dilated (nonischemic) cardiomyopathy LBBB left bundle branch block L-S lateral-septal LV left ventricular dp/dt max peak positive left ventricular pressure NYHA New York Heart Association PATH-CHF Pacing Therapies for Congestive Heart Failure study RV right ventricular provement of LV asynchrony was quantified with tissue Doppler imaging from the apical views (9), but this technique is limited to the study of longitudinal axis motion. Two-dimensional Fourier phase imaging may quantify wall motion asynchrony in the radial direction and has been used to assess LV asynergy in coronary artery disease (CAD) (10). We hypothesized that the degree of radial ventricular asynchrony in patients with heart failure and ventricular conduction delay predicts the magnitude of contractile function improvement with CRT. To that end, we have evaluated a new phase analysis technique to quantify regional wall motion synchrony from endocardial border contours generated semi-automatically from twodimensional echocardiographic ventricular images. METHODS Patients. The PAcing THerapies in Congestive Heart Failure (PATH-CHF) trial is a prospective, multicenter, single-blinded, cross-over study conducted in Europe; it included 42 patients with ischemic and nonischemic cardiomyopathy, with a QRS width 120 ms and PR interval 150 ms. All patients had to be in stable New York Heart Association (NYHA) functional class III or IV, without the need for intravenous inotropic drugs. A detailed study design has been reported elsewhere (11). Because this study was initiated before availability of dedicated CRT systems, all patients received a biventricular (BV) pacing system with two separately implanted dual-chamber (DDD) pacemakers, an apical RV lead and an epicardial LV lead, implanted during a limited thoracotomy of the LV free wall. Biventricular pacing was obtained by programming one device in DDD mode and the second device in a ventricular triggered mode. This configuration enabled noninvasive testing of different pacing sites. The echocardiographic results obtained in the patient group were compared with those of a control group of 10 healthy individuals with a normal PR interval and QRS width. Invasive optimization. During implantation, invasive hemodynamic testing was performed using the FLEXSTIM system (Guidant Corp., St. Paul, Minnesota) (11), with repeated measurements of dp/dt max at various atrioventricular (AV) delays and pacing sites (RV, LV and BV) tested in random order in the VDD mode. The response to pacing was expressed as the percent increase in dp/dt max (% dp/ dt max ), compared with no pacing. Details of the invasive optimization procedure and pacemaker implantation have been described elsewhere (3). Evaluation of invasive variables was performed with no knowledge of the results of the echocardiographic analysis. Echocardiographic analysis. For baseline evaluation, the transthoracic echocardiograms of 34 patients were analyzed if there was sufficient image quality for complete endocardial border delineation. Studies were recorded in the left lateral supine position at rest in the week before implantation of the CRT system. To minimize the influence of relative motion of the heart, only echocardiographic recordings that could be obtained in respiratory hold and with a stable transducer position were included. Fundamental imaging was used in the majority of the baseline examinations (n 26); harmonic imaging was used whenever it was available to the study center (n 8). At the first follow-up visit, four weeks after implantation, echocardiographic recordings were made with temporary reprogramming of the CRT system to no pacing (OFF) and to RV, LV, and BV VDD pacing in random order. For each individual, the AV delay was programmed close to the optimal setting, as determined by acute invasive testing during implantation and kept constant for each pacing mode. Valid echocardiographic images from four-week follow-up were available for analysis in 16 patients. Two patients were excluded from the study because of their high pacing thresholds; two patients had sudden cardiac death; and 14 patients were excluded because they had technically inadequate echocardiographic recordings in at least one tested pacing mode. All examinations were recorded on S-VHS videotape and digitized for wall motion analysis with the CMS echo-analysis system (Medis, Leiden, Netherlands) (12) at the responsible core center (University Hospital, Aachen, Germany). Quantification of ventricular asynchrony. During the cardiac cycle, each region of the ventricular endocardial wall undergoes a cycle of inward and outward displacement. Each regional displacement cycle can be represented as a curve of displacement plotted over time from the start to the end of a cardiac cycle interval. Because these displacement curves are periodic, they can be analyzed in the frequency domain to quantify the phase relationship between curves independent of the displacement magnitude and heart rate. Each regional displacement curve is modeled as a wave with a period equal to the cardiac cycle interval, which, mathematically, is the fundamental frequency in Fourier analysis. The time at which the center of this wave occurs during the cardiac cycle interval is a function of the fundamental frequency phase angle ( ). It is near 180 when centered in the middle of the cycle, 0 to 180 if shifted earlier and 180 to 360 if shifted later. Inverted and triphasic displacement

538 Breithardt et al. JACC Vol. 40, No. 3, 2002 Quantification of Ventricular Asynchrony in CRT August 7, 2002:536 45 Figure 1. (A) End-diastolic image (apical four-chamber view) with manually drawn left ventricular (LV) endocardial contour tracing. (B) Left ventricular wall motion displacement for 100 endocardial segments determined by using the centerline method. (C) Septal (dashed line) and lateral (solid line) wall motion averaged for 40 septal and lateral segments and 3 to 7 cardiac cycles and displayed as displacement (mm) over time (s). curves (e.g., with paradoxical septal wall motion) have phase angles near the end (360 ) or start (0 ) of the cycle. With this method, the magnitude of synchrony between two regional displacement curves is calculated by the difference between their respective phase angles. Phase differences near 0 indicate near-perfect synchrony, whereas a difference of 180 defines maximal asynchrony. All wall motion analyses were performed with no knowledge of the invasive hemodynamic test results and the patients clinical characteristics. The pacing mode was marked on videotape for identification. The CMS semiautomatic border detection software was used to delineate and track the LV endocardial wall motion in sequential frames of digitized images from the apical four-chamber view. Analysis of the apical two-chamber view was not feasible in the majority of patients because of incomplete border delineation (in most cases, the anterior wall). End diastole was demarcated by the frame in which the mitral valve first began to close; end systole was demarcated by the frame in which the mitral valve first began to open. Wall motion contours (Fig. 1A) were manually drawn in the first systolic and diastolic frames of each cardiac cycle, and the CMS software automatically generated intermediate frame contours, which were manually adjusted as necessary. For each CRT mode, endocardial motion was tracked through three to seven cardiac cycles verified to be in normal sinus rhythm by the concurrent surface electrocardiographic recording. Regional endocardial displacement was calculated for each cardiac cycle automatically by the CMS software, using the centerline method for 100 equally spaced segments on the LV wall motion contours (Fig. 1B). This method has been shown to reduce interobserver variability in the delineation of endocardial boundaries (12). Forty segments from the basal septum toward the apex and 40 segments from the basal lateral wall toward the apex were averaged for calculation of septal and lateral regional displacement curves (Fig. 1B and 1C). Regional displacement curves were ensemble-averaged over three to seven cardiac cycles using the first systolic frame as the fiducial marker. Each curve was offset to zero displacement at the start of each cycle. Before phase analysis, the average regional displacement curves were smoothed with a threeframe moving-average filter. Septal and lateral displacement phases were defined by the phase angle of the fundamental frequency of the Fourier transform computed over the cardiac cycle regional displacement curve: (D ): tan 1 D,sin D,cos. This phase angle was computed with the discrete frame data, using the inner product of the regional displacement curve and orthogonal sine and cosine curves of the cardiac

JACC Vol. 40, No. 3, 2002 August 7, 2002:536 45 Breithardt et al. Quantification of Ventricular Asynchrony in CRT 539 cycle interval length. Septal displacement curves exhibiting paradoxical negative systolic displacement that yielded a very small phase angle ( 60 ) due to the 360 modulus were adjusted to 360 S. Lateral and septal (L and S) phase relationships were measured by the difference between the lateral ( L ) and septal ( S ) phase angles: LS L S. The absolute value of LS was used as an orderindependent measure of synchrony: LS L S. Statistics. Continuous data are expressed in the text as the mean value SD and in the figures as the mean value SEM. To evaluate and compare the effects of RV, LV and BV pacing with no pacing treatments on the hemodynamic and echocardiographic measurements of each individual, we used a general linear model (analysis of variance [ANOVA]) accounting for all treatment variations being tested in each patient. To compare measurements among the control and L-S phase type groups, we used independent-samples ANOVA. For both ANOVAs, the Tukey correction was used to correct for type I error inflation introduced by testing multiple hypotheses. An unpaired t test was used to compare characteristics of analyzed and excluded patient groups and to compare measurements from patients with CAD and dilated cardiomyopathy (DCM). Statistical analyses were made with SAS version 8.2 (SAS Institute, Cary, North Carolina). The reproducibility of endocardial border delineation and phase angle measurements was assessed in 10 randomly selected baseline examinations as the mean difference between two independent measurements performed on different occasions by one observer (intraobserver variability) and between two independent observers (interobserver variability). The results were expressed as the percentage of the first measurement ( SD) and also as the percentage of 180 ( SD), based on the fact that two measurements cannot differ by 180 over the 360 cycle. RESULTS At baseline, the 34 patients (mean age 59 6 years; 19 men and 15 women) presented, in the majority of cases, with NYHA functional class III (n 33), LBBB (n 32) and nonischemic DCM (n 24). The mean QRS width was 176 34 ms; the mean PR interval was 211 38 ms; and the LV ejection fraction was significantly reduced (mean 21 6%). The mean % dp/dt max with optimized CRT during invasive testing was 7.6 7.7% with RV pacing, 19.2 15.6% (p 0.001 vs. RV) with LV pacing and 17.8 14.5% (p 0.001 vs. RV) with BV pacing. The mean intrinsic AV interval for the patient sample was 221 38 ms, and the average programmed AV interval during follow-up CRT testing and echocardiographic recording was 107 28 ms. All individuals in the control group presented with a normal echocardiographic LV ejection fraction of 60%. The 16 patients studied at the first follow-up visit at four weeks was comparable to the 18 excluded patients in terms of age (59 6 vs. 60 6 years, p NS), baseline QRS (172 32 vs. 179 36 ms, p NS), LS (82 37 vs. 87 54 ), baseline dp/dt max (600 161 vs. 527 83 mm Hg, p NS) and % dp/dt max (21 14% vs. 19 17%, p NS). All patients were receiving stable pharmacologic therapy from baseline to four-week follow-up, except for one patient who began beta-blocker therapy just before the four-week follow-up. Baseline L-S phase relationships. All control subjects were characterized by monophasic lateral and septal displacements with LS 25, which we defined as nearsynchronous phase (Fig. 2A). Three distinct types of L-S phase relationships were retrospectively identified in the 34 patients studied at baseline. A type 1 pattern, similar to the observed pattern in the control group, was apparent in four patients and was characterized by monophasic lateral and septal displacements with LS 25 (mean LS 5 6 ) (Fig. 2B). A type 2 pattern was defined by a septal phase preceding the lateral phase by 25 with either monophasic or biphasic septal displacement (Fig. 2C), which was observed in 17 patients (mean LS 77 33 ). Thirteen patients showed a type 3 pattern (mean LS 115 33 ) with a late septal phase (Fig. 2D). This pattern was usually associated with triphasic or inverted monophasic septal displacement. Table 1 summarizes the distribution of L-S phase types and the corresponding patient characteristics. The baseline dp/dt max tended to be highest and the QRS duration shortest in type 1 patients, and this group showed the least benefit from pacing, as measured by mean % dp/dt max with CRT. Table 2 compares the noninvasive measures of LV asynchrony (QRS width and LS ) with the individual hemodynamic responses to CRT. None of the four type 1 patients had improved dp/dt max with CRT, although one had a QRS duration of 153 ms and baseline dp/dt max 500 mm Hg/s (patient no. 4). In contrast, although two type 2 patients had a QRS duration 130 ms (patient nos. 5 and 6), ventricular preexcitation due to BV CRT led to 11% to 17% increases in dp/dt max. Three patients did not have improved dp/dt max with CRT, despite pronounced type 3 asynchrony (patient nos. 22 to 24). We observed a unimodal relationship between the dp/ dt max response at the best possible CRT setting in each patient and their baseline LS (Fig. 3). Patients who exhibited large increases in dp/dt max at the best CRT setting tended to have a large positive or negative baseline LS value, corresponding to a large degree of L-S asynchrony. Patients who exhibited small increases in dp/dt max at the best CRT setting tended to have a small baseline LS value, corresponding to more synchronous L-S displacement. No significant differences were observed between patients with DCM and CAD, although those with DCM tended to show a slightly larger QRS width at baseline (183 32 vs. 160 34 ms, p 0.07), a higher LS (93 46 vs. 66 43, p 0.13) and a larger hemodynamic response to CRT (mean % LVdP/dt max 22 15 vs. 16 15%, p 0.07).

540 Breithardt et al. JACC Vol. 40, No. 3, 2002 Quantification of Ventricular Asynchrony in CRT August 7, 2002:536 45 Figure 2. Examples of the observed types of wall motion patterns. Consecutive cardiac cycles (3 to 7) were averaged to show wall motion for lateral (solid line) and septal (dashed line) segments, as displacement over time. Effects of CRT on L-S synchrony. Sixteen patients were studied four weeks after implantation to test the early effects of CRT on mean L-S synchrony, as measured by the change in LS during reprogramming of the pacemakers. During intrinsic conduction (OFF), the mean LS was 104 41, which decreased to 86 45 with RV CRT (p 0.14 vs. OFF), to 71 50 with LV CRT (mean difference 33, 95% confidence interval [CI] 54 to 11, p 0.001 vs. OFF) and to 66 42 with BV CRT (mean difference 38, 95% CI 59 to 17, p 0.001 vs. OFF). Percent synchrony improvement with each CRT mode was associated with proportional percent increases in dp/dt max (Fig. 4). Compared with RV pacing, LV and BV pacing resulted in significantly larger increases in dp/dt max (p 0.001) and tended to have larger differences in synchrony improvement (p 0.14 RV vs. LV, and p 0.12 RV vs. BV). Type 2 patients (n 8) exhibited a significant LS decrease from 84 26 (OFF) to 36 26 at the best CRT mode (p 0.001 vs. OFF by the paired t test) (Fig. 5A and 5B). In contrast, type 3 patients (n 8) showed less change, with a nonsignificant LS decrease from 123 46 (OFF) to 105 41 at the best CRT mode (p NS by the paired t test). However, CRT eliminated or reversed the early septal inward movement in type 3 patients (Fig. 5C and 5D). The correlation between LS and dp/dt max changes with CRT was significant for type 2 patients (n 8) who Table 1. Lateral-Septal Phase Relationship Types Control Subjects (n 10) Type 1 Patients (n 4) Type 2 Patients (n 17) Type 3 Patients (n 13) L 148 19 183 35 202 34 * 184 24 * S 167 21 178 32 125 37 * 303 35 * LS 19 19 5 6 77 33 * 119 31 * QRS duration (ms) 78 9 134 14* 186 33* 176 30* Baseline dp/dt max (mm Hg/s) ND 692 310 532 148 558 154 dp/dt max with optimized CRT ND 2 1% 26 14% 18 15% *p 0.05 vs. control subjects. p 0.05 vs. type 1 patients. p 0.05 vs. type 2 patients. Data are presented as the mean value SD. CRT cardiac resynchronization therapy; ( )dp/dt max (change in) peak positive left ventricular pressure; ND not done.

JACC Vol. 40, No. 3, 2002 August 7, 2002:536 45 Breithardt et al. Quantification of Ventricular Asynchrony in CRT 541 Table 2. Noninvasive Measures of Asynchrony and Individual Hemodynamic Response to Cardiac Resynchronization Therapy Patient* No. Type Baseline QRS Duration (ms) Baseline LS ( ) Baseline dp/dt max (mm Hg/s) Best CRT dp/dt max (%) Best CRT Mode 1 1 123 10.62 882.76 2.62 NR 2 1 124 7.08 411.92 0.42 NR 3 1 135 3.40 1,028.15 2.19 NR 4 1 153 5.11 445.82 1.03 NR 5 2 128 64.68 674.02 16.86 BV 6 2 130 38.62 780.47 11.01 BV 7 2 160 37.97 597.49 28.67 BV 8 2 166 47.16 625.73 2.64 NR 9 2 169 44.49 434.46 27.91 LV 10 2 172 51.07 811.57 9.15 LV 11 2 176 105.14 470.20 39.62 LV 12 2 184 55.83 702.77 39.55 BV 13 2 191 115.55 396.71 16.16 BV 14 2 193 65.71 361.66 41.62 BV 15 2 193 109.17 487.61 32.01 LV 16 2 196 45.84 309.38 53.81 LV 17 2 198 91.66 519.63 10.29 LV 18 2 202 61.20 400.47 19.55 LV 19 2 210 127.04 522.32 25.00 LV 20 2 221 125.88 551.14 36.26 LV 21 2 268 118.53 405.94 34.92 LV 22 3 124 102.30 767.32 0.21 NR 23 3 128 120.52 847.35 0.55 NR 24 3 148 178.52 582.22 1.16 NR 25 3 168 104.95 590.91 20.35 LV 26 3 172 112.50 443.44 12.05 LV 27 3 178 61.84 775.26 19.45 LV 28 3 178 133.76 588.02 24.47 BV 29 3 178 139.26 356.59 43.05 LV 30 3 181 76.86 486.32 6.64 BV 31 3 193 153.37 478.41 19.67 LV 32 3 194 99.76 424.16 46.87 LV 33 3 215 121.90 502.99 25.94 LV 34 3 228 138.46 407.97 16.09 LV *Patients were sorted by their QRS duration in each type group and assigned identifying numbers. Patients with right bundle branch block. BV biventricular; LV left ventricular; NR no significant response in dp/dt max with 5% change from OFF; other abbreviations as in Table 1. had LV and BV CRT, but failed to reach significance for those who had RV CRT (Fig. 6). No significant correlation between LS and dp/dt max was observed in type 3 patients. Reproducibility. We found a good reproducibility of phase angle analysis: 8 11 for repeated measurements (intraobserver variability) (adjusted to 180 : 5 6%) and 15 11 for two independent observers (interobserver variability) (adjusted to 180 : 8 6%). DISCUSSION The study demonstrates a unique echocardiographic method for quantifying LV mechanical wall motion synchrony, which can be used to predict a hemodynamic contractile function benefit from CRT. Increased dp/dt max due to CRT was directly associated with improved LV mechanical synchrony, as measured by a reduction in the absolute L-S phase angle LS in type 2 patients with delayed lateral wall inward movement. Also, this is the first study to noninvasively assess, by two-dimensional echocardiography, the effects of different CRT stimulation sites on LV mechanical synchrony and compare them with invasively measured hemodynamic responses. Both LV and BV CRT significantly improved LV L-S synchrony, whereas less improvement was observed with RV CRT. This is consistent with previous reports that LV and BV CRT increase dp/dt max to a much greater extent than RV CRT (3,13). It is well established that basal contractile function is depressed in patients with heart failure with DCM due to alterations in the contractile machinery within each myofibril (14) and in the extracellular matrix (15). Other studies suggest that in addition to altered molecular contractility, another cause of lowered contractile function is reduced cooperation among contracting myofibrils due to asynchronous LV contraction (16). Contractile cooperation, as we shall call this proposed second dimension of contractile

542 Breithardt et al. JACC Vol. 40, No. 3, 2002 Quantification of Ventricular Asynchrony in CRT August 7, 2002:536 45 Figure 3. Baseline LS correlated with the best improvement in contractile function with cardiac resynchronization therapy. Data points are fitted by regression analysis with a second-order polynomial forced to pass through the origin (0,0): % LVdP/dt max 0.098 (baseline LS ) 0.0016 (baseline LS ) 2. The correlation coefficient was calculated for a regression through the origin, and significance was tested with analysis of variance (R 2 0.54, p 0.001). Vertical dashed lines separate the different types of wall motion patterns. LVdP/dt max peak positive left ventricular pressure. function, may be reduced in patients with heart failure by abnormal ventricular conduction delays (2). Analogously, it can be reduced by RV pacing that advances contraction of the paced region relative to normally synchronous LV regions (16). Early activated regions contract against low chamber pressures but waste energy by prestretching the opposing, nonstimulated regions. Conversely, the late activated, excessively preloaded regions contract against a higher wall stress. This reduced contractile cooperation reduces overall cardiac efficiency and increases myocardial energy demands (17). Our wall motion phase results suggest that LV CRT can advance the start of delayed lateral wall contractions to improve synchrony with early septal wall contractions, and BV CRT can stimulate simultaneous Figure 4. Improvement in LS (open bars) and LVdP/dt max peak positive left ventricular pressure (shaded bars) displayed as the percent change from no pacing (OFF) for every cardiac resynchronization (CRT) mode (RV, LV and BV). Data are presented as the mean value SEM. n 16. *p 0.001 vs. RV. BV biventricular; LV left ventricular; RV right ventricular. lateral and septal wall contractions, both of which improve contractile cooperation, as indicated by increased dp/dt max. Influence of the pacing site. Experimental data on normal dogs show that RV-only pacing creates early septal and late lateral LV wall contractions; LV-only pacing creates early lateral and late septal LV wall contractions; and simultaneous BV pacing minimizes asynchrony (16,18). The asynchronous contractions with LV-only pacing were observed at very short AV delays that prevented any fusion with intrinsic ventricular activation. In contrast, our results, predominantly in patients with LBBB, demonstrate that L-S wall motion is synchronized nearly as well by LV CRT as by BV CRT. Our hypothesis for this paradox is that resynchronization with LV CRT requires an optimal AV delay, such that the paced lateral wall activation combines with the intrinsic AV-conducted septal wall activation. Even with BV CRT, the resulting wall motion patterns are likely to be a complex function of the two paced wave fronts and intrinsic AV-conducted activation, so that maximal resynchronization depends on an optimal AV delay. This importance of an optimized AV delay might also explain the conflicting results of Kerwin et al. (7), who analyzed multiple-gated blood-pool scintigraphic images and found an increase in left intraventricular dyssynchrony with BV pacing at a fixed AV delay. Another paradox is that apical RV CRT is able to improve synchrony in patients with delayed LV lateral wall movement, although to a lesser degree than LV and BV CRT. Earlier studies reported that apical RV pacing can improve dp/dt max and aortic pulse pressure in patients with heart failure and LBBB (3). Xiao et al. (19) showed that the LV electromechanical delay was shorter with apical RV pacing compared with intrinsic activation with LBBB. Thus, apical RV pacing with an optimized AV delay must be able to preexcite at least some areas of the LV, compared with intrinsic activation, but to a lesser extent than LV and BV CRT and, on average, with less hemodynamic benefit. Predictive value of baseline mechanical asynchrony for hemodynamic improvement. The multiple L-S phase relationships we observed suggest that patients with heart failure with comparable ventricular conduction delays can have markedly different underlying mechanical abnormalities. Patients with a QRS duration 150 ms could exhibit near-synchronous L-S displacements (type 1), very delayed lateral displacements (type 2) or paradoxical septal motion (type 3). The type 1 pattern with a prolonged QRS complex probably results from a symmetrical conduction delay across the septal and lateral regions. In this group, CRT did not result in improved hemodynamic function, despite the presence of wide QRS complexes and a very low baseline dp/dt max value. Type 2 patients presented with delayed lateral wall motion and exhibited the most benefit from CRT. The acute reduction in LS in type 2 patients correlated well with the rise in dp/dt max, as documented during invasive testing. This included patients with QRS complexes 155 ms and baseline dp/dt max 700 mm Hg/s,

JACC Vol. 40, No. 3, 2002 August 7, 2002:536 45 Breithardt et al. Quantification of Ventricular Asynchrony in CRT 543 Figure 5. Effect of cardiac resynchronization (CRT) on regional displacement curves. (A) Regional asynchrony in a type 2 patient with delayed inward movement of the lateral wall (solid line) in relation to the septum (dashed line). (B) Synchronized lateral and septal inward movement by biventricular (BV) CRT. (C) Triphasic septal movement pattern in type 3. Early septal inward movement (arrow) precedes the lateral wall, followed by septal outward movement during lateral wall inward movement. This corresponds to the previously described paradoxical septal motion in left bundle branch block. (D) Early septal inward movement is no longer evident during BV CRT. OFF no pacing. who, by these criteria, would have been predicted to be acute hemodynamic CRT nonresponders, according to earlier studies (8). Lateral-septal synchrony improved in the majority of type 3 patients, but the change in LS was not proportional to the percent increase in dp/dt max. It is possible that the fundamental frequency phase analysis is not sensitive enough to adequately quantify changes in the complex biphasic and triphasic septal wall motion patterns in type 3 patients; in which case, higher order frequency components might provide additional predictive information. Three patients (Table 2, patient nos. 22 to 24) did not show improved dp/dt max with CRT, despite pronounced type 3 asynchrony. These exceptions might represent a limitation of our L-S phase measure to predict an acute CRT response for type 3 patients. It might be speculated that in these type 3 patients, the lateral wall does not correspond to the site with the longest electromechanical delay, and that the additional evaluation of anteriorposterior synchrony in the two-chamber view might have improved the results. This group might also represent patients with suboptimal lead locations, who might have benefited from CRT with alternative stimulation sites: in Patient no. 24 the lead was placed near to the LV base in a posterolateral position and in Patient no. 23 the lead was located on the anterior wall. Both positions have been associated with suboptimal acute effects of CRT (20). Clinical implications. Several echocardiographic measures have been proposed to screen or optimize CRT for patients with heart failure, such as transmitral and aortic Doppler echocardiography, three-dimensional echocardiography and tissue Doppler imaging. Sogaard et al. (21) recently demonstrated that assessment of longitudinal function by tissue Doppler echocardiography is able to predict improvement in the LV ejection fraction with CRT, and Yu et al. (9) reported that CRT reduces the regional difference in myocardial peak systolic velocities. However, the new metric LS is the first to show a direct relationship between invasively measured hemodynamic improvement with CRT and LV mechanical synchrony assessed by analysis of radial wall motion. We suggest it might provide a noninvasive screening method for patients with heart failure, so that those likely to have increased contractile function with CRT can be selected and so that CRT after implantation can be optimized. Baseline asynchrony indicated by LS 25 predicts a contractile function benefit from CRT. For patients with type 2 L-S phase patterns, the magnitude of

544 Breithardt et al. JACC Vol. 40, No. 3, 2002 Quantification of Ventricular Asynchrony in CRT August 7, 2002:536 45 Figure 6. The change in LS with cardiac resynchronization therapy (CRT) predicted the improvement in peak positive left ventricular pressure (LVdP/dt max ) in type 2 patients. A weak, nonsignificant correlation was observed for right ventricular (RV) CRT (A). The strongest correlation was observed for left ventricular (LV) CRT (B), and the effects during BV CRT correlated moderately to hemodynamic improvement (C). OFF no pacing. LS reduction with CRT correlates to the invasively measured increase in dp/dt max. Limitations. This study is limited to the echocardiographic prediction of an acute hemodynamic response, and it is unclear how these predictions will extend to long-term clinical benefit. Echocardiographic analysis was limited to the apical four-chamber view. It is possible that although CRT pacing at lateral LV sites resynchronizes L-S wall motion, it may not resynchronize or could delay anteroposterior wall motion. However, previous results with tissue Doppler echocardiography suggest that the effects of CRT are, to a large degree, confined to the interventricular septum and the inferoposterior and lateral walls (21). We were able to identify a subgroup (type 2) in whom the apical four-chamber view provided important information about the associated hemodynamic improvement. Harmonic imaging was only used in a minority of patients studied; it is expected to improve endocardial border delineation in segments with suboptimal visualization of the endocardium, thereby possibly improving measurement variability and enabling evaluation of additional segments. The programmed AV delays were defined individually in every patient according to the best invasive hemodynamic results and kept constant in every CRT mode. The effect of AV interaction was not systematically evaluated. Echocardiographic testing with the different CRT modes was performed after four weeks of CRT, which could have altered basal wall motion patterns; a different magnitude of CRT effects might be obtained if measurements are made immediately after device implantation. Endocardial border displacement methods cannot distinguish between active and passive wall motion and do not take into account changes in myocardial wall thickness. Therefore, changes in L-S phase relationships could be due to changes in RV-LV transseptal pressure gradients, as well as changes in intraventricular synchrony. It should be noted that the same limitation applies to measurement of regional myocardial velocities by tissue Doppler imaging. Strain rate imaging with calculation of regional myocardial velocity gradients might overcome that limitation in the future. Translational and rotational movement of the heart in relation to the transducer is an inherent problem of all imaging techniques and might influence the regional phase angle values of the walls. However, the opposing walls will be affected to a similar degree, and the relative differences of L-S phase relationships will be less affected (10). We tried to minimize these effects during the examination (respiratory hold, stable transducer position) and during computation of regional phase shifts (averaging of multiple cycles with offset to zero displacement at the start of each cycle). The temporal resolution of wall motion was limited to video recording frame rates of 25 frames/s. Integration on echocardiographic work stations with direct on-line contouration could improve temporal resolution and, thereby, accuracy and reproducibility. Conclusions. Despite the promising results of CRT on both acute hemodynamic performance and long-term functional status, the selection of suitable patients is still ill defined. A ventricular conduction delay, as measured by the

JACC Vol. 40, No. 3, 2002 August 7, 2002:536 45 Breithardt et al. Quantification of Ventricular Asynchrony in CRT 545 QRS duration, only weakly predicts the expected hemodynamic benefit with CRT (8). Echocardiographic phase analysis of radial endocardial wall motion demonstrates that optimized CRT restores LV synchrony by normalizing septal wall movement and advancing lateral wall activation in relation to the septum. Quantitative echocardiography is able to identify patients likely to have an acute hemodynamic benefit and provides a noninvasive alternative for identification of possible CRT candidates. The magnitude of resynchronization is proportional to the acute contractile function response in a defined subgroup. This may help to optimize CRT during follow-up, particularly in type 2 patients. Further studies are warranted to evaluate the relationship between the degree of LV resynchronization and its long-term benefit on exercise capacity, functional mitral regurgitation and LV reverse remodeling. Acknowledgments We thank Mr. Jeng Mah (Guidant Corp., St. Paul, Minnesota) for assisting with the statistical analysis. The investigators also deeply acknowledge the help and support of all nurses and colleagues at their institutions. Reprint requests and correspondence: Dr. Christoph Stellbrink, Medizinische Klinik I der RWTH Aachen, Pauwelsstrasse 30, D-52057 Aachen, Germany. E-mail: cstellbrink@ukaachen.de. REFERENCES 1. Grines CL, Bashore TM, Boudoulas H, Olson S, Shafer P, Wooley CF. Functional abnormalities in isolated left bundle branch block: the effect of interventricular asynchrony. Circulation 1989;79:845 53. 2. Fried AG, Parker AB, Newton GE, Parker JD. Electrical and hemodynamic correlates of the maximal rate of pressure increase in the human left ventricle. J Card Fail 1999;5:8 16. 3. Auricchio A, Stellbrink C, Block M, et al. Effect of pacing chamber and atrioventricular delay on acute systolic function of paced patients with congestive heart failure. Circulation 1999;99:2993 3001. 4. Breithardt OA, Stellbrink C, Franke A, et al. Acute effects of cardiac resynchronization therapy on left ventricular Doppler indices in patients with congestive heart failure. Am Heart J 2002;143:34 44. 5. Stellbrink C, Breithardt OA, Franke A, et al. Impact of cardiac resynchronization therapy using hemodynamically optimized pacing on left ventricular remodeling in patients with congestive heart failure and ventricular conduction disturbances. J Am Coll Cardiol 2001;38: 1957 65. 6. Le Rest C, Couturier O, Turzo A, et al. Use of left ventricular pacing in heart failure: evaluation by gated blood pool imaging. J Nucl Cardiol 1999;6:651 6. 7. Kerwin WF, Botvinick EH, O Connel JW, et al. Ventricular contraction abnormalities in dilated cardiomyopathy: effect of biventricular pacing to correct interventricular dyssynchrony. J Am Coll Cardiol 2000;35:1221 7. 8. Nelson GS, Curry CW, Wyman BT, et al. Predictors of systolic augmentation from left ventricular preexcitation in patients with dilated cardiomyopathy and intraventricular conduction delay. Circulation 2000;101:2703 9. 9. Yu CM, Chau E, Sanderson JE, et al. Tissue Doppler echocardiographic evidence of reverse remodeling and improved synchronicity by simultaneously delaying regional contraction after biventricular pacing therapy in heart failure. Circulation 2002;105:438 45. 10. Hansen A, Krueger C, Hardt SE, Haass M, Kuecherer HF. Echocardiographic quantification of left ventricular asynergy in coronary artery disease with Fourier phase imaging. Int J Card Imaging 2001;17:81 8. 11. Auricchio A, Stellbrink C, Sack S, et al. The PAcing THerapies for Congestive Heart Failure (PATH-CHF) study: rationale, design, and endpoints of a prospective randomized multicenter study. Am J Cardiol 1999;83:130D 5D. 12. Bosch JG, Savalle LH, van Burken G, Reiber JH. Evaluation of a semiautomatic contour detection approach in sequences of short-axis two-dimensional echocardiographic images. J Am Soc Echocardiogr 1995;8:810 21. 13. Kass DA, Chen CH, Curry C, et al. Improved left ventricular mechanics from acute VDD pacing in patients with dilated cardiomyopathy and ventricular conduction delay. Circulation 1999;99:1567 73. 14. Gomez AM, Valdivia HH, Cheng H, et al. Defective excitationcontraction coupling in experimental cardiac hypertrophy and heart failure. Science 1997;276:800 6. 15. Li YY, Feng Y, McTiernan CF, et al. Downregulation of matrix metalloproteinases and reduction in collagen damage in the failing human heart after support with left ventricular assist devices. Circulation 2001;104:1147 52. 16. Prinzen FW, Hunter WC, Wyman BT, McVeigh ER. Mapping of regional myocardial strain and work during ventricular pacing: experimental study using magnetic resonance imaging tagging. J Am Coll Cardiol 1999;33:1735 42. 17. Nelson GS, Berger RD, Fetics BJ, et al. Left ventricular or biventricular pacing improves cardiac function at diminished energy cost in patients with dilated cardiomyopathy and left bundle-branch block. Circulation 2000;102:3053 9. 18. van Oosterhout MF, Prinzen FW, Arts T, et al. Asynchronous electrical activation induces asymmetrical hypertrophy of the left ventricular wall. Circulation 1998;98:588 95. 19. Xiao HB, Brecker SJ, Gibson DG. Differing effects of right ventricular pacing and left bundle branch block on left ventricular function. Br Heart J 1993;69:166 73. 20. Butter C, Auricchio A, Stellbrink C, et al. Effect of resynchronization therapy stimulation site on the systolic function of heart failure patients. Circulation 2001;104:3026 9. 21. Sogaard P, Kim WY, Jensen HK, et al. Impact of acute biventricular pacing on left ventricular performance and volumes in patients with severe heart failure: a tissue Doppler and three-dimensional echocardiographic study. Cardiology 2001;95:173 82.