Uncoupling at the Atrial Level

The right atrium and left atrium are activated nearly simultaneously (within 50-80 milliseconds) during sinus rhythm. Preferential sites of interatrial conduction exist at the posterior-superior interatrial septum (Bachmann’s bundle region), fossa ovalis, and coronary sinus ostium.^1,2 Significant interatrial conduction delays (up to 200 msec or greater) can occur in myopathic atria. Similar conduction delays can be also be induced, or exacerbated, by right atrial (RA) pacing. Delayed left atrial (LA) contraction can disrupt optimal left-sided atrioventricular (AV) coupling. Severe atrial decoupling and delayed LA contraction reverses the left-sided AV contraction sequence, resulting in atrial transport block.³ This causes increased LA pressures, retrograde flow in the pulmonary veins, and counterphysiologic neurohormonal responses termed pseudo–pacemaker syndrome.^4–6

Uncoupling at the Atrioventricular Level

Optimal AV coupling contributes to ventricular pump function through Starling’s law. The normal intrinsic atrioventricular interval (iAVI) results in atrial contraction just before the preejection (isovolumetric) period of ventricular contraction that maximizes left ventricular (LV) filling (LV end-diastolic pressure, or preload) and cardiac output. Appropriately timed AV coupling maintains a low mean atrial pressure, increases LV filling time, maximizes preload and pump function (Starling mechanism), and positions the mitral valve for closure during the isovolumetric phase of ventricular systole (Fig. 31-1).

Figure 31-1 Schematic representation of optimal atrioventricular (AV) coupling.

Atrial contraction (A) occurs just before isovolumetric contraction (ic); ir, isovolumetric relaxation. Passive diastolic left ventricular (LV) filling (E) is unimpeded, and total diastolic filling time (E + A) is maximized. Left atrial pressure (blue line) remains low during diastole, LV pressure (orange line) is high during ejection.

Delayed AV coupling resulting from AV conduction delay displaces atrial contraction earlier in diastole (Fig. 31-2). Atrial contraction occurs during ventricular filling, which shortens diastolic filling time, and in extreme situations, occurs immediately after or coincident with the preceding ventricular contraction. This situation is aggravated by ventricular conduction delay, which increases ventricular ejection time and delays diastolic filling, increasing the probability of collision with atrial contraction. Atrial contraction before completion of venous return reduces preload stretching of the left ventricle, which reduces ventricular volume and contractile force. Delayed AV coupling reduces diastolic filling time and may result in diastolic mitral regurgitation when elevated LV end-diastolic pressure exceeds LA pressure. Partial closure of the mitral valve may also occur, further shortening diastolic filling time.

Figure 31-2 Schematic of hemodynamic effects of AV decoupling on LV pump function.

Same arrangement as in Figure 31-1. Atrial contraction truncates passive diastolic filling (E-A fusion). A late diastolic left ventricular–left atrial (LV-LA) pressure gradient before mitral valve closure results in diastolic mitral regurgitation. LV filling is reduced and pump function declines (reduced LV pressure during ejection).

Delayed AV coupling may be worsened or induced by atrial pacing. This is a consequence of atrial capture latency, increased atrial conduction time resulting from altered activation, and increased AV nodal conduction time from altered atrial input. These effects of atrial pacing are enhanced by increased heart rates during atrial pacing and autonomic changes during sleep.⁷

Severe AV decoupling coexists with ventriculoatrial (VA) coupling. In extreme conditions LA contraction occurs immediately after, within, or preceding the LV contraction, causing reversal of the left-sided AV contraction sequence,³ which initiates counterphysiologic circulatory and neurohumoral reflexes. These derangements (pseudo–pacemaker syndrome) have been rarely described during marked first-degree AV block and atrial pacing.⁴ The identical pathophysiology occurs during AV synchronous ventricular pacing when LV contraction timing is too early, resulting in LA transport block (Fig. 31-3). This situation is corrected by proper timing of ventricular contraction with pacing (see Atrioventricular Optimization).

Figure 31-3 Hemodynamic effects of short A-V interval on LV pump function.

Same arrangement as in Figures 31-1 and 31-2. Ventricular contraction triggered by premature pacing stimulus (arrow) truncates atrial emptying (atrial transport block). Atrial pressure increases, LV filling decreases, and pump function declines (reduced LV pressure during ejection).

Atrial fibrillation disrupts atrial coupling (atrial desynchronization) and eliminates AV coupling (AV uncoupling).

Effects of Asynchrony on Left Ventricular Mechanics and Structure

Regions that are activated early also start to contract early. Accordingly, in RVA pacing and LBBB, earliest and latest sites of segmental LV activation correlate well with time to peak systolic velocity by Doppler imaging¹⁹ and strain by tagged magnetic resonance imaging (MRI),^9,26–28 providing evidence that ventricular conduction delay is responsible for heterogeneous mechanical performance (asynchrony). The mechanical effect of asynchronous electrical activation is dramatic, because the various regions differ not only in the time of onset of contraction, but also in the pattern of contraction (Fig. 31-4). Contraction of early-activated myocardium is energetically inefficient because LV pressure is low and ejection has not begun. Instead, stretching of remote regions not yet activated absorbs the energy generated by the early-activated regions. This stretching further delays shortening of these late activation regions and increases their force of local contraction by virtue of the Frank-Starling mechanism (locally enhanced preload). Vigorous late systolic contraction at delayed sites occurs against high LV cavity pressures (locally enhanced afterload) and imposes loading on the earlier activated regions, which now undergo systolic paradoxical stretch.²⁹ This reciprocated stretching of regions within the LV wall causes a less effective and energetically efficient contraction.³⁰

Figure 31-4 Effects of synchronous (left panel) and asynchronous (right panel) ventricular activation on LV and aortic pressure (A), regional strain (B), ventricular conduction (C), electrocardiographic tracings), RV and LV pressure signals (D), and pressure-volume loops (E). Asynchronous contraction causes reduction in ejection time and slows rates of rise and fall of LV pressure and aortic pressure and increases duration of isovolumetric contraction (ic) and relaxation (ir) (A). Onset of LV shortening (strain) is regionally delayed (negative deflection of curve) (B). Asynchronous electromechanical activation induces increased QRSd (C) and interventricular/intraventricular activation times (D), which instantaneously reduce stroke volume, stroke work, dP/dt_max, and dP/dt_min, indicating an acute reduction in pump function (E).

(Modified from Sweeney MO, Prinzen FW: Ventricular pump function and pacing: physiological and clinical integration. Circ Arrhythmia Electrophysiol 1:127-139, 2008.)

The hemodynamic consequences of the asynchronous LV contraction are reductions in contractility and relaxation. These changes occur immediately on initiating RVA pacing in humans and animals,^11,31 and induction of LBBB in animals¹² (see Fig. 31-4). The loss of pump function is indicated by decreases in stroke volume, stroke work, and slower rate of rise of LV pressure (Fig. 31-5). Moreover, the LV end-systolic pressure-volume relationship shifts rightward, indicating that the left ventricle operates at a larger volume in order to recruit the Frank-Starling mechanism.¹¹ Premature relaxation in early-activated regions, and delayed contraction in others also causes abnormal relaxation,¹¹ as expressed as slower rate of fall of LV pressure, increase in the relaxation time constant tau, and decrease of E-wave velocity amplitude on Doppler echocardiograms.³² These changes also lead to prolongation of the isovolumetric contraction and relaxation times, which are characteristic for asynchronous hearts.³³ The prolongation of the isovolumetric times occurs mainly at the expense of the diastolic filling time, leading to reduced preload.

Figure 31-5 Left bundle branch block (LBBB)–induced asynchrony causes immediate reduction in LV contractility.

A, Left ventricular (LV) endocardial activation maps. Inner circle represents LV apex and outer circle represents LV base. S, A, L, and P indicate the septum, anterior, lateral, and posterior walls, respectively. Arrows indicate activation delay vectors, the amplitude of which reflects the degree of intraventricular asynchrony (intra VA). B, Surface electrocardiogram (ECG) tracings, where QRS duration indicates total ventricular asynchrony (total VA). C, Right ventricular (RV) and LV pressure signals normalized to reveal pressure differences (interventricular asynchrony, inter VA). D, Pressure-volume (PV) loops. Loop area is stroke work, loop width is stroke volume.

During normal ventricular conduction (top row, pre-LBBB) intraVA and interVA are minimized, QRSd is normal and LV contractility is greater (larger PV loop). After induction of LBBB (bottom row, post-LBBB), intra VA, inter VA, and total VA are increased, which immediately reduces LV pump function (smaller PV loop indicating declines in stroke work, stroke volume, and LV pressure generation).

(Modified from Vernooy K et al: Left bundle branch block induces ventricular remodeling and functional septal hypoperfusion. Eur Heart J 26:91-98, 2005.)

The redistribution of the mechanical load within the ventricular wall also leads to reduction of regional myocardial perfusion and oxygen consumption near the RVA pacing site^30,34,35 and in the septum during LBBB.^36,37 Such perfusion defects and wall motion abnormalities have been demonstrated in up to 70% of patients with angiographically normal coronary arteries exposed to chronic RVA pacing.^38–41 These defects are reversible on cessation of RVA pacing^39,41 and after biventricular pacing.^36,37,42 Accordingly, such perfusion “deficits” do not necessarily indicate coronary heart disease, because they may simply express the regional differences in myocardial workload.

Similar to the situation after myocardial infarction, acute loss of pump function initiates compensatory responses (Figs. 31-6 and 31-7). Some of these responses, after a certain time and or degree of asynchrony, may result in further impairment of pump function and clinical heart failure. There are various triggers for these “remodeling” processes. As for other conditions of hemodynamic overload, RVA pacing and LBBB lead to stimulation of the sympathetic system, resulting in elevated myocardial catecholamine levels,⁴³ and activation of the renin-angiotensin-aldosterone system. Regional differences in stretch and mechanical load heterogeneity are most likely important stimuli for remodeling processes. Chronic asynchronous LV activation results in regional and global structural changes indicated by asymmetric hypertrophy, increased end-diastolic volume, and reduced ejection fraction^36,44 (see Fig. 31-6), as well as locally different molecular abnormalities, including reductions in sarcoplasmic reticulum calcium-ATPase and phospholamban.⁴⁵ Even stronger regional differences in gene expression are found in failing hearts with conduction abnormalities.⁴⁶ Dystrophic calcifications, disorganized mitochondria, and myofibrillar cellular disarray⁴⁷ have been described with RVA pacing, especially in immature hearts.

Figure 31-6 Left bundle branch block (LBBB).

*P, 0.05 compared to baseline. †Significant change over time by ANOVA. A, Effects of induced LBBB on electromechanics. Dashed lines on surface electrocardiogram (ECG) indicate earliest and latest left ventricular (LV) endocardial activation times. Bull’s-eye plots indicate septal (S), posterior (P), lateral (L), and anterior (A) LV wall, where outer ring indicates LV base and inner ring indicates LV apex. Activation is color-coded where lighter shades indicate later activation times. During normal ventricular conduction (Baseline), QRSd is narrow and LV electromechanical activation is simultaneous throughout. During LBBB, QRS is prolonged and LV electrical and mechanical activation is delayed with lateral wall latest. B-D, Change in LV end-diastolic volume (increase), LV septal-lateral wall mass ratio (decrease, indicating asymmetrical hypertrophy), and ejection fraction (decrease) after 16 weeks of LBBB in a canine model. The purple and red lines in B and D indicate the time course for two individual dogs that were most and least sensitive to the adverse electromechanical effects of LBBB.

(Modified from Vernooy K et al: Left bundle branch block induces ventricular remodeling and functional septal hypoperfusion. Eur Heart J 26:91-98, 2005.)

Figure 31-7 Mechanisms of ventricular remodeling and progressive reduction in pump function during mechanical dyssynchrony induced by ventricular conduction delay (LBBB, RVA pacing). For details, see text.

(Modified from Sweeney MO, Prinzen FW: Ventricular pump function and pacing: physiological and clinical integration. Circ Arrhythmia Electrophysiol 1:127-139, 2008).

Early signs of cellular and molecular adaptation to disturbed ventricular activation are manifest as reduction in ejection fraction (EF) within the first week of RVA pacing. In patients with normal EF, Nahlawi et al.³¹ showed an immediate drop in EF of about 7%, followed by another 7% drop during the subsequent week. Suppression of RVA pacing returned EF to baseline values, but only after several days.⁴⁸ Also, ventricular repolarization abnormalities have been observed within hours of RVA pacing. Alterations in potassium and calcium channels likely play a role in these phenomena.

Mechanistic Basis for Development of Asynchronous Heart Failure Caused by Ventricular Conduction Delay

As mentioned, synchronous activation consistently leads to acute and long-term adverse effects on cardiac pump function, which create a vicious cycle of deterioration. Accordingly, the pacing-induced ventricular conduction delay has been linked to increased risks of atrial fibrillation, heart failure, ventricular tachyarrhythmia, and death in large, randomized clinical trials of pacemakers and ICDs.^49–56 Similar risks have been reported for LBBB and cardiac morbidity and mortality.⁵⁷

Multiple factors may contribute to heart failure associated with ventricular asynchrony. At least three candidate factors are readily identified: (1) reduced pump function caused by asynchronous contraction (dyssynchrony), (2) adverse remodeling caused by long-term dyssynchrony, (3) left-sided AV desynchronization, and (4) functional mitral regurgitation (fMR). The first three factors are discussed earlier. The fMR frequently accompanies pacing-induced or spontaneous asynchronous heart failure and has multiple mechanisms.^24,25,58,59 Assuming the anterior mitral leaflet is structurally normal, the immediate cause of fMR is delayed activation of the LV posterior “apparatus” caused by ventricular conduction delay.^25,60 This apparatus consists of the posterior leaflet and annulus, the LV summit it sits on, the chordae, the papillary muscles, and the LV wall from which they emanate. Rapid activation of the posterior apparatus requires normal ventricular conduction.

Early closure of the mitral valve is arguably the most important accomplishment of the entire conduction system because it facilitates Starling’s law; that is, it permits isovolumetric systole to occur when the ventricle is as full as it can be given the filling conditions. Delayed ventricular conduction prevents mitral valve coaptation and induces fMR during isovolumetric contraction. Ventricular conduction delay instantaneously reduces the transmitral pressure gradient (systolic LV pressure–LA pressure difference) or mitral valve closing force, by reducing contractility (↓LV dP/dt). Chronically, volumetric remodeling contributes to fMR by further reductions in contractility (closing forces) and increasing papillary muscle tethering forces^59,61,62 (Fig. 31-8). Tethering strongly depends on alterations in ventricular shape because the tethering forces that act on the mitral leaflets are higher in dilated, more spherical ventricles. Ventricular dilation and increased chamber sphericity increase the distance between the papillary muscles to the enlarged mitral annulus as well as to each other, restricting leaflet motion and increasing the force needed for effective mitral valve closure. Under these conditions, the mitral regurgitant orifice area will be largely determined by the phasic changes in transmitral pressure.

Figure 31-8 Effects of augmented mitral valve leaflet tethering in dilated cardiomyopathy.

Functional mitral regurgitation (fMR) results from an imbalance between the closing and tethering forces that act on the mitral leaflets. When the mitral valve is tethered, mitral regurgitant orifice area is largely determined by phasic changes in transmitral pressure (TMP). Reductions in LV pump function (↓dP/dt) results in decreased rate of LV (and thus transmitral) pressure, which will increase MR severity because of the impaired closing force.

(Modified from Otsuji Y, et al: Restricted diastolic opening of the mitral leaflets in patients with left ventricular dysfunction: evidence for increased valve tethering. J Am Coll Cardiol 32:398-404, 1998.)

There is some intriguing evidence that LBBB may be the primary cause of DCM in some patients. Progressive LV enlargement after the emergence of LBBB unaccompanied by any other explanation has been documented in case series.⁶³ Furthermore, complete recovery of ventricular function shortly after initiation of biventricular pacing in “hyperresponders” suggests that LBBB may be a primary and reversible trigger of DCM in some patients.^64,65

Several studies also showed that LBBB is associated with increased mortality, with greater impact in sicker patient populations. The relative risk associated with the presence of LBBB varied between 1.5 and 2.0, even after adjustment for covariates.⁵⁷ The only study rejecting the association of LBBB with increased mortality drew this conclusion after adjustments for LV EF as a covariable.⁶⁶ However, the conclusion that LBBB is not an independent predictor of mortality after the adjustment for EF can be understood because LBBB directly reduces EF because of its acute and chronic effect on LV pump function (see earlier).

Improved Atrioventricular Mechanics from Optimal AV Resynchronization

When LV contraction timing is delayed, even when AV conduction times are normal, the hemodynamic consequences are similar to a prolonged AV interval, as shown in Figure 31-2. Left-sided AV decoupling, due to LV conduction delay and/or AV conduction delay, is corrected with LV pacing. Optimal AV timing improves LV pump function, but the primary mechanism of improved ventricular mechanics during CRT is reduction in ventricular conduction delay. AV timing considerations during CRT are considered in detail later.

Improved Ventricular Mechanics from Reduced Ventricular Conduction Delay

Reduction of intraventricular delay during atrial-synchronous LV or biventricular pacing has an immediate positive effect on ventricular mechanics and is the dominant therapy effect of CRT. Instantaneous improvement in pump function is indicated by increases in LV +dP/dt_max, stroke volume, stroke work, arterial pulse pressure and peak systolic pressure,^67,68 and reduced end-systolic volume (Fig. 31-9). Moreover, unlike the inotropic effects of dobutamine, systolic augmentation with ventricular resynchronization increases efficiency of conversion of myocardial oxygen consumption to mechanical work.⁶⁹ This positive contractile response demonstrates a modestly positive correlation with increasing baseline QRSd^67–69 and is strongly correlated with baseline mechanical dyssynchrony.^69,70 Conversely, neither improved contractility nor reverse volumetric remodeling requires QRS narrowing during LV or biventricular pacing.^68,71,72 However, unlike the inotropic effects of dobutamine, systolic augmentation with ventricular resynchronization does not increase detrimental myocardial oxygen consumption. These acute improvements in ventricular mechanics are maintained chronically and are accompanied by reverse volumetric remodeling of the LV (see later).⁷³ Abrupt discontinuation of LV pacing results in immediate reduction in indices of improved contractility and regression of reverse remodeling over several weeks,^73,74 as well as recurrence of fMR.²⁵

Figure 31-9 Pressure-volume loops during LBBB and LV pacing.

Correction of regional mechanical delay results in immediate reduction in left ventricular (LV) volumes (leftward shift of pressure-volume relationship) and increases in stroke work (larger loop), stroke volume (width of loop), and dP/dt_max. CRT, Cardiac resynchronization therapy.

(Modified from Verbeek XA, Vernooy K, Peschar M: Intra-ventricular resynchronization for optimal left ventricular function during pacing in experimental left bundle branch block. J Am Coll Cardiol 42:558-567, 2003.)

Maximally effective ventricular resynchronization occurs at the midpoint of the interventricular (V-V) interval (Fig. 31-10), which is the conduction time from RV to LV pacing site during native activation. A single optimum AV delay will achieve maximally effective ventricular resynchronization by advancing LV activation sufficiently to minimize the V-V interval. The AV delay and the V-V interval have an interactive relationship. The effect of AV delay on acute hemodynamic response to CRT is best understood in terms of the effect on ventricular resynchronization during biventricular (BiV) or left univentricular pacing (Fig. 31-11). During LBBB, the right ventricle is activated by right bundle branch conduction (RBBc) before the left ventricle, and the time difference between RV and LV activation is the interventricular conduction delay.¹³ BiV pacing at short atrioventricular delay (AVD) results in complete replacement of intrinsic activation, allowing complete control of chamber timing with sequential ventricular stimulation. Intermediate AVDs result in fusion among RBBc, RV paced activation, and LV paced activation. Long AVDs result in “pure” fusion between RBBc and LV paced activation since LV preexcitation is still possible because of interventricular conduction delay.^13,75,76 Modification of the AVD alone can therefore achieve (1) BiV paced fusion, (2) BiV paced fusion with sequential ventricular stimulation (LV first), and (3) “pure” RBBc-LV pacing fusion. Therefore, LV preexcitation is possible across a range of AVDs because LV time is delayed relative to RV time. RV pacing is not necessary to achieve fusion, and AVD can be used to time the relative prematurity of LV preexcitation. At very short AVDs, this requires sequential VV timing. At very long AVDs, this occurs spontaneously because of the native V-V interval.

Figure 31-10 Relationship between intraVA and interVA under various LV pacing conditions.

At long atrioventricular delay (AVD), ventricular conduction delay persists (large intraVA and interVA). At very short AVD, large intraVA and interVA persist because complete reversal of LV activation has occurred. At the optimal AVD, intraVA is minimized and interVA resides at the midpoint of the interVA during either extreme condition, which yields the maximum improvement in pump function.

(Modified from Verbeek XA, Vernooy K, Peschar M: Intra-ventricular resynchronization for optimal left ventricular function during pacing in experimental left bundle branch block. J Am Coll Cardiol 42:558-567, 2003; and Vernooy K, et al: Calculation of effective VV interval facilitates optimization of AV delay and VV interval in cardiac resynchronization therapy. Heart Rhythm 4:75-82, 2006.)

Figure 31-11 Top, Interaction between atrioventricular delay (AVD) and V-V interval to determine LV activation fusion. T-RV is time to intrinsic RV activation (RA sense to RV sense, or P-R interval). During short AVD pacing (AVD < T-RV), biventricular pacing completely replaces native activation (surrogate fusion), and relative timing of ventricular activation is determined by sequential ventricular stimulation (RV first, simultaneous, LV first). When AVD = T-RV, RV pacing occurs simultaneously with intrinsic conduction, resulting in a combination of RV pacing and intrinsic conduction, whereas left ventricle is still activated earlier than would have occurred due to LBBB. When AVD > T-RV, LV-only pacing occurs with true fusion of native and paced activation. Bottom, Epicardial isochrone maps are shown during LV-only pacing at increasing pacemaker A-V intervals (pAVIs). Activation times relative to QRS onset are color-coded with red earliest and blue latest; framed numbers indicate earliest and latest activated sites. Asterisk, LV stimulation site. Black lines indicate line/region of conduction block, which influences activation wavefront propagation (see Fig. 31-77). A, pAVI = 0 msec. Note sequential ventricular activation (LV→RV): LV is activated earliest, and latest activation occurs at posterobasal right ventricle. B to C, Progressive fusion of paced LV activation with intrinsic LV activation via RBB as pAVI is increased from 65 to 86 msec. Note also progressively early RV activation from RBB conduction. D, Maximum ventricular activation wavefront fusion at pAVI = 173 msec. Note right and left ventricles are simultaneously activated (Esyn, electrical synchrony, = −18 msec).

(Modified from Jia P et al: Electrocardiographic imaging of cardiac resynchronization therapy in heart failure: observation of variable electrophysiologic responses. Heart Rhythm 3:296-310, 2006.)

Early acute hemodynamic studies demonstrated that the relationship between AVD and LV dP/dt_max in patients with an acute improvement in contractile response to simultaneous BiV pacing is positive and unimodal, with a peak effect at approximately 50% of the native P-R interval, resulting in complete replacement of native ventricular conduction with paced activation^67,68,77 (Fig. 31-12). This dose-response relationship between AVD and acute contractile response shown is interpreted as follows. The magnitude of the acute contractile response depends on the magnitude of baseline asynchrony (ventricular conduction delay, QRSd) and the AVD, which titrates how much intrinsic ventricular activation is replaced by pacing activation. Patients with greater ventricular conduction delay (QRS >150 msec) benefit from near-complete replacement of intrinsic activation with BiV or LV pacing (i.e., short AVDs). Patients with less ventricular conduction delay (QRS <150 msec) benefit from limited replacement of intrinsic activation (e.g., long AVDs). Typically, programmed short AVDs of 100 to 120 msec, although near-optimal for some patients, may result in decreased contractile function in patients with less baseline asynchrony (e.g., QRS <150 msec) because even BiV pacing induces asynchrony.⁹ LV dP/dt_max increased 15% to 45% across a narrow range of AVDs and declined at very short AVDs (truncated filling times) or very long AVDs (inadequate correction of LV conduction delay). This maximum improvement in pump function (LV +dP/dt_max and stroke volume) occurs at minimal intraventricular asynchrony and unchanged LV end-diastolic pressure (preload).¹³ This indicates that the acute contractile response is explained by ventricular resynchronization, not by better filling, and that the dose of resynchronization is titrated by the AVD.

Figure 31-12 Interaction between atrioventricular delay (AVD), magnitude of LV conduction delay, and acute improvement in contractility during CRT.

Mean (±SEM) change in LV dP/dt_max when BiV pacing at various AVDs in patients with QRS duration greater than 150 msec (diamonds) or 120 to 150 msec (triangles). The AVDs tested were fixed percentages of the intrinsic A-V interval (iAVI) determined individually for each patient. Acute hemodynamic responders (type I) demonstrate improvement across a broad range of AVDs, whereas hemodynamic nonresponders (type II) may worsen because of very short AVDs.

(Modified from Auricchio A et al: Effect of pacing chamber and atrioventricular delay on acute systolic function of paced patients with congestive heart failure. Circulation 99:2993-3001, 1999.)

The best ventricular resynchronization occurs when the wavefronts from the RV pacing site (or originating from RBBc) and the LV pacing site collide halfway.⁷⁸ Experimental evidence suggests that optimal hemodynamic effect can be achieved using LV pacing at certain AVDs as well as using simultaneous and sequential BiV pacing because fusion of electrical wavefronts can be achieved by either approach^13,79 (see Fig. 31-10). Although this and other experimental studies show a clear mechanism of action for optimizing the interaction between AV and VV timing, confusion surrounds this area, and a clinical advantage of patient-specific timing optimization has not been convincingly demonstrated in clinical trials^80–82 (see later).

Reverse Volumetric Left Ventricular Remodeling

Mechanical resynchronization reduces mechanical load heterogeneity throughout the left ventricle. Sustained improvement in ventricular mechanics results in regression of adverse LV remodeling, termed reverse remodeling. This is indicated by reductions in LV volumes and mass, reduced mitral orifice size, regression of asymmetric hypertrophy, and increased EF.^73,83–90 The magnitude of the reduction in end-systolic volume ranges between 10% and 30%,^83,87,91,92 which is equivalent or greater than the effects of beta-adrenergic blockers and angiotensin-converting enzyme (ACE) inhibitors. These hemodynamic effects are chronically sustained assuming LV pacing is continuously maintained.⁷³ Although improvement in New York Heart Association (NYHA) functional class, quality of life, and 6-minute walk tests may be confounded by placebo effects and reporting bias, reverse remodeling is a completely objective measure. More importantly, it appears to be related to better survival.^92,93

Evidence of greater baseline electromechanical asynchrony and acute mechanical resynchronization appears to be necessary for reverse remodeling to occur, and the greater the reduction in asynchrony, the higher the probability of remodeling.^72,90,94 Likewise, absence of electromechanical resynchronization acutely eliminates the possibility of chronic reverse remodeling.^72,90 In canine hearts with induced LBBB, 8 weeks of CRT was sufficient to reverse almost completely the about 25% LV cavity dilation and asymmetric LV hypertrophy induced by dyssynchrony to pre-LBBB baseline values.⁴² This observation has been replicated in clinical studies where in some cases, reverse remodeling results in complete normalization of LV volumes and EF.⁶⁴ This suggests that ventricular conduction delay may be the primary cause of DCM in some patients,^63–65 as it was to a more moderate degree in the LBBB dogs.

Reduction in Functional Mitral Regurgitation

In many patients, CRT reduces fMR, which likely accounts for immediate reduction in symptoms in advance of possible reverse volumetric remodeling. This can be attributed to the following complex factors:

Figure 31-13 One mechanism for acute reduction in severity of functional mitral regurgitation during CRT.

Schematic representation of the relationship between the increase in transmitral pressure (TMP) and the decrease in effective mitral valve regurgitant orifice area (EROA). Top, Left bundle branch block (LBBB). Left ventricular (LV) contractility is low, indicated by a slow rate of rise and delayed peak in LV pressure. This results in delayed and reduced TMP such that EROA remains large throughout systole. Bottom, Cardiac resynchronization therapy (CRT). Acute improvement in contractility, indicated by rapid rate of rise and early peak of LV pressure, results in higher TMP earlier in systole, which reduces EROA. Shaded area represents the time in systole during which EROA is below 50% of its initial value. Reduction in regurgitant V wave preserves TMP.

(Modified from Breithardt OA, et al: Acute effects of cardiac resynchronization therapy on functional mitral regurgitation in advanced systolic heart failure. J Am Coll Cardiol 203:765-770, 2003.)

Figure 31-14 Effect of earlier delivery of posterior apparatus on functional mitral regurgitation (fMR).

Echocardiographic strain images from the four-chamber view and two-chamber view (top), with corresponding time-strain plots from sites adjacent to papillary muscles before and after CRT (bottom). Baseline plots demonstrate delayed peak strain occurring in the anterolateral papillary muscle site compared with the posteromedial papillary muscle site. CRT results in time alignment of peak strain at papillary muscles (top). Time to peak systolic strain is color-coded, with lines representing isochrones of mechanical activation times at 50-msec intervals. The X indicates sites of lead placement, and the arrow indicates the direction of the propagating mechanical activation. Time to peak strain of sites adjacent to anterolateral (AL P) and posteromedial (PM P) papillary muscles are shown. A decrease in interpapillary muscle time delay was associated with decreased fMR.

(Modified from Kanzaki H, et al: A mechanism for immediate reduction in mitral regurgitation after cardiac resynchronization therapy: insights from mechanical activation strain mapping. J Am Coll Cardiol 44:1619-1625, 2004.)

Factors 1 to 3 are acute effects and unrelated to geometrical changes (reverse remodeling). Factors 2 to 4 are the primary determinants of reduction in fMR and are directly related to reduction in ventricular conduction delay. It is not well understood what part of the benefit of CRT is caused by reduction in fMR versus resynchronization of contraction directly, because few studies quantitatively measure regurgitant flow.

Improved Heart Rate Profile

Ventricular resynchronization favorably modifies autonomic balance, as evidenced by reduced mean heart rate and increased heart rate variability, which correlate with improved functional capacity, reverse remodeling, and reduced mortality.⁹⁵

Leads and Electrodes: Non–Independently Programmable Ventricular Polarity Configurations

Transvenous and epicardial LV pacing leads may be either unipolar or bipolar, although the former dominates current applications. Multiple ventricular pacing polarity configurations are therefore possible. Because programmed polarity settings are common to both ventricular leads, and the type (bipolar or unipolar) of these leads may not be the same, the following considerations apply.

In a dual-bipolar configuration, both lead tips are the active electrodes (cathodes) and the rings are the common (nonstimulating) anode. However, the type of ventricular leads implanted defines the pacing/sensing vector (Figs. 31-15 and 31-16). With two unipolar leads, the bipolar setting results in no pacing or sensing. If both leads are bipolar, both rings act as the common electrode. If one lead is bipolar (RV) and the other lead is unipolar (typically LV), the ring on the bipolar lead acts as the common electrode (nonstimulating anode). This configuration results in shared-ring bipolar pacing and sensing. This hybrid bipolar/unipolar stimulation configuration (“dual cathodal”) is employed in most contemporary CRT pacing systems.

Figure 31-15 Biventricular pacing configurations.

(From Barold SS, Stroobandt RX, Sinnaeve AF: Cardiac pacemakers step by step: an illustrated guide. Malden, Mass, 2004, Blackwell.)

Figure 31-16 Leads and electrodes for biventricular pacing.

(From Barold SS, Stroobandt RX, Sinnaeve AF: Cardiac pacemakers step by step: an illustrated guide. Malden, Mass, 2004, Blackwell.)

In a dual-unipolar configuration, the lead tips are the active electrodes; the noninsulated device case is the common electrode. This polarity configuration is infrequently used in CRT pacing systems and is not feasible in CRT-D systems because of the concerns regarding ventricular oversensing associated with the unipolar pacing stimulus.

Pulse Generators

Conventional dual-chamber pulse generators or specially designed multisite pacing pulse generators may be used for CRT applications (Fig. 31-17). A conventional dual-chamber pulse generator is well suited for CRT in patients with permanent atrial fibrillation. In this situation, the ventricular port is used for the RV lead and the atrial port is used for the LV lead. This permits programming of independent outputs and ventricular-ventricular timing by manipulation of the AV delay. The programming mode can be either DDD/R or DVI/R (see later). A conventional dual-chamber pulse generator can also be used for atrial-synchronous BiV pacing. The single ventricular output must be divided to provide simultaneous stimulation of the RV and LV (dual cathodal system with parallel outputs). This is achieved with a Y adapter and results in simultaneous RV and LV sensing, which may result in ventricular double-counting and loss of CRT, or pacemaker inhibition in the case of LV lead dislodgment into the coronary sinus with sensing of atrial activity.⁹⁶

Figure 31-17 Pulse generators for biventricular pacing.

(From Barold SS, Stroobandt RX, Sinnaeve AF: Cardiac pacemakers step by step: an illustrated guide. Malden, Mass, 2004, Blackwell.)

First-generation multisite pacing pulse generators similarly provide a single ventricular output for simultaneous RV and LV stimulation. However, two separate ventricular channels internally connect in parallel. This connection is made for both the lead tip and ring connections and eliminates the need for a Y adapter. This configuration still provides simultaneous RV and LV sensing with associated limitations.

Second-generation multisite pacing pulse generators have independent ventricular ports. Each ventricular lead therefore has separate sensing and output circuits. This arrangement permits optimal programming of outputs and time delay between RV and LV stimulation for each patient. It also eliminates the potential complications of BiV sensing.

Pacing Modes

It is axiomatic that for maximal delivery of CRT, ventricular pacing must be continuous.

The DDD mode (atrial and ventricular pacing/sensing) guarantees AV synchrony by synchronizing ventricular pacing to all atrial events, except during episodes of atrial tachycardia or atrial fibrillation. However, DDD mode increases the probability of atrial pacing (depending on programmed lower rate limit) that may alter the left-sided AV timing relationship because of interatrial conduction time and atrial pacing latency.

The VDD mode (atrial sensing only, ventricular pacing and sensing) guarantees the absence of atrial pacing and synchronizes all atrial events to ventricular pacing at the programmed AV delay. However, if the sinus rate is below the lower programmed rate limit, AV synchrony is lost because the VDD mode is operationally VVI (ventricular-only sensing and pacing).

Although conventional dual-chamber pacing systems are not designed for BiV pacing and generally do not allow programming of an AV delay of zero, or near zero, they are being increasingly used with their shortest AV delay (0-30 msec) for CRT in patients with permanent atrial fibrillation. The advantages include programming flexibility, elimination of the Y adapter (required for conventional VVIR devices), protection against far-field sensing of atrial activity (an inherent risk of dual cathodal devices with simultaneous sensing from both ventricles), and cost. When a conventional dual-chamber pacemaker is used for CRT, the LV lead is usually connected to the atrial port and the RV lead to the ventricular port. This provides LV stimulation before RV activation (LV preexcitation) and protection against ventricular asystole related to oversensing of far-field atrial activity, when the LV lead is dislodged toward the AV groove.

The DVIR mode is ideally suited for this application. The DVIR mode (committed atrial pacing, ventricular pacing and sensing) behaves similar to the VVIR mode, except that there are always two closely coupled independent ventricular stimuli, thereby facilitating comprehensive evaluation of RV and LV pacing and sensing performance. The DVIR mode also provides absolute protection against far-field sensing of atrial activity in case of LV lead dislodgment, because no sensing occurs on the “atrial” (LV) lead in the DVIR mode.

Determining LV and RV Capture: Importance of Electrocardiography

The 12-lead electrocardiogram (ECG) is essential to ascertain RV and LV capture during follow-up of CRT systems without separately programmable ventricular outputs. It is recognized that at least six distinct 12-lead ventricular activation patterns may be seen during threshold determination: (1) intrinsic rhythm during loss of RV and LV capture or pacing inhibition (native QRS), (2) isolated RV stimulation, (3) isolated LV stimulation, (4) BiV stimulation with complete replacement of native ventricular activation by paced activation, (5) BiV stimulation with fusion between native activation (RBB conduction) and paced activation, and (6) BiV stimulation with anodal capture. Further, fusion between paced and native activation may result in a wide range of unique global activation patterns on the 12-lead ECG (see later).

Ventricular pacing thresholds should ideally be performed independently and in the VVI mode at a rate exceeding the prevailing ventricular rate so as to obtain continuous ventricular capture uncontaminated by fusion with native activation. Alternately, thresholds can be performed in the VDD or DDD mode at very short AV delays (e.g., 50% of P-R interval) to ensure full ventricular capture without fusion. In older devices without separately programmable ventricular outputs, RV and LV capture can only be determined by ECG analysis during common ventricular voltage decrement. In this situation it is advisable to initiate threshold determinations at maximum output (voltage and pulse duration) because a significant differential often exists in capture thresholds between RV (lower) and LV (higher). Knowledgeable analysis of the ventricular activation sequence using the 12-lead surface ECG reliably establishes univentricular capture.

Monochamber RV apical (RVA) pacing generates right (R) → left (L) activation in the frontal plane and anterior (A) → posterior (P) activation in the horizontal plane, resembling LBBB activation (Fig. 31-18). Consequently, mean QRS frontal plane axis is usually left superior and is infrequently right superior. Pacing from the midsuperior RV septum also generates right (R) → left (L) activation in the frontal plane and anterior (A) → posterior (P) activation in the horizontal plane, but the mean QRS frontal plane axis is typically left inferior. Monochamber LV pacing produces a variety of activation sequences depending on stimulation site and conduction blocks. A few generalizations are possible. Posterior wall stimulation usually generates an L → R, P → A activation, and mean QRS frontal plane axis is usually right superior, less often right inferior. A detailed discussion of ventricular activation sequences during monochamber LV and BiV pacing is provided later. Figure 31-19 shows a typical example of ventricular activation during monochamber RVA versus monochamber LV capture threshold determination.

Figure 31-18 Mean frontal plane QRS axis during various conditions of ventricular pacing.

(From Barold SS, Stroobandt RX, Sinnaeve AF: Cardiac pacemakers step by step: an illustrated guide. Malden, Mass, 2004, Blackwell.)

Figure 31-19 Typical ventricular activation sequences during monochamber RV and LV pacing threshold determination.

Top, Right ventricular (RV) apex pacing. Activation is R → L, A → P; frontal plane axis is left superior. Bottom, Left ventricular (LV) posterior wall pacing. Activation is L → R, P → A; frontal plane axis is right superior.

The majority of current CRT-P/CRT-D systems use either dual cathodal or dedicated bipolar pacing configuration. Anodal capture refers to the situation when myocardial capture occurs at the RV anode in a dual cathodal arrangement. This could theoretically occur in isolation with the LV cathode, but most often occurs with both RV and LV cathodes, and is referred to as “triple-site” pacing, an unfortunate misnomer; a more accurate descriptor would be “capture at three sites during biventricular pacing.” Anodal capture is more common at high-voltage outputs and with true bipolar RV leads because of the small surface area and higher current density of the ring electrode, as opposed to the larger surface area and lower current density of the coil electrode in integrated bipolar leads.

Anodal capture results in a distinct change in activation pattern compared to BiV pacing that can only be appreciated on the 12-lead ECG (Fig. 31-20). The change in QRS morphology related to loss of anodal capture as voltage output is decremented during a temporary threshold test using a single ECG lead may be misinterpreted as loss of LV capture and may result in erroneous overestimation of the LV threshold.

Figure 31-20 Effect of anodal capture on ventricular activation sequence during simultaneous biventricular (BiV) pacing.

Same patient as in Figure 31-19. RV apex and LV posterior wall capture thresholds are less than 1 V. Top, Simultaneous BiV pacing when LV output is 5 V (5 times > threshold) resulting in anodal capture. Bottom, Simultaneous BiV pacing when LV output is 2 V and no anodal capture. Note change in activation sequence. During high-LV-output BiV pacing, anodal capture is indicated by attenuation of monophasic R wave in V1 and loss of R (QS) in V2 (horizontal plane) and attenuation of QS in lead I and aVL (frontal plane). QRS duration is shorter. During lower-output BiV pacing, loss of anodal capture is indicated by dominant QS in I, aVL, and dominant R (V1) and RS (V2). QRS duration is longer.

The physiologic consequences of anodal capture are uncertain. One study demonstrated that anodal capture might be advantageous during CRT by counteracting the regional activation delay located at the inferior wall of the left ventricle and improving regional measures of intraventricular dyssynchrony.⁹⁷

Automatic Adjustment of Left Ventricular Pacing Output

Left ventricular capture management (LVCM) provides automatic adjustment of LV pacing output voltage based on periodic assessment of LV capture (Medtronic). As with all automatic threshold adjusting protocols, the intent of LVCM is to deliver continuous LV pacing at the minimum output of voltage that yields 100% capture. The primary purpose of LVCM is battery conservation. This is clinically relevant because LV voltage thresholds are typically higher than RV thresholds, rendering a single voltage output for both sites inefficient. An ancillary value is the possibility of overcoming phrenic stimulation when the voltage differential between phrenic nerve capture (higher) and LV capture (lower) is sufficiently differentiated. In this situation, LV output voltage can be automatically “capped” beneath the phrenic nerve capture voltage without compromising LV capture. Provision of long-term threshold behavior as a dedicated diagnostic measure might reduce follow-up time by eliminating the need for in-office real-time determinations, and is particularly useful during remote surveillance.

When LVCM is enabled, the device automatically monitors the pacing amplitude threshold at periodic intervals, nominally 01:00 am daily. The minimum amplitude that consistently results in ventricular capture (threshold) is stored and reported. If LVCM is programmed to “adaptive,” LV output voltage is automatically reprogrammed toward a selectable maximum output voltage defined by the desired “maximum amplitude safety margin” and reported. Alternately, if LVCM is programmed to “monitor,” LV output is not adjusted, although data regarding threshold determinations are recorded.

Operating Details

Unlike automatic RV pacing threshold adjusting algorithms in conventional pacemakers, LVCM does not use the evoked response to determine ventricular capture. Rather, LVCM determines LV capture by recording RV sensed events in response to LV monochamber pacing. This requires timing methods to ascertain whether RV sensed events during monochamber LV pacing are caused by native AV conduction, ventricular premature beats, electromagnetic interference (EMI), or other extraneous conditions. RV sensed events outside the “expected” timing window for native AV conduction, or loss of RV sensing during monochamber LV pacing, is used to indicate loss of LV capture (e.g., threshold voltage).

The LVCM operation consists of four stages: (1) ventricular rate and stability check, (2) LV-RV conduction check, (3) AV conduction check, and (4) LV pacing threshold search (LVPTS). The order of these stages is not arbitrary. Ventricular rhythm must be stable (<200 msec R-R interval variability) and rate less than about 90 beats per minute (bpm) for 12 cycles. The reason for this is the need to ensure stable AV conduction timing and LV-RV conduction timing during atrial-synchronous monochamber LV pacing in tracking modes and/or 100% overdrive ventricular pacing during atrial fibrillation (nontracking modes).

The LV (paced) → RV (sensed) conduction time is determined during cycles of up to eight (atrial synchronous) monoventricular pacing sequences (Fig. 31-21). To enhance accurate measurement of unidirectional ventricular conduction time, this test is initiated approximately 15 bpm above the ventricular rate measured during the rate and stability check. In dual-chamber modes, the pacemaker atrioventricular interval (pAVI) is shortened to 30 msec. These two adjustments are intended to guarantee complete replacement of AV and VV conduction with paced activation. Additionally, the blanking period on the RV sense amplifier is minimized to enhance RV sensing. LV-RV conduction is performed with LV output at the prevailing programmed setting. If RV sensing does not occur in response to a LV-only test stimulus, a backup BV pacing stimulus with LV at maximum amplitude occurs on the following cycle, and LV output remains at maximum amplitude except for single-test cycles.

Figure 31-21 Left ventricular capture management (LVCM) operation.

Top, Surface ECG, event markers, right ventricular electrogram (RV EGM). Successful LV-RV and AV conduction checks. During overdrive atrial pacing and LV monochamber pacing, the LV-RV interval is 180 msec (left). During AV conduction check (right) PAV is set to 180 msec + 80 msec (260 msec). Note that this example was obtained from a prototype RAMware download; in the market-released application, the PAV would be set to 180 + 60 (240 msec); see text for details. Absence of intrinsic AV conduction within this timing window is confirmed, satisfying the condition for LV Pacing Threshold Search (LVPTS). Bottom, Surface ECG, event markers for LVPTS. The expected LV-RV conduction time is 180 msec, which sets the capture detection window 150 to 200 msec (30 msec earlier—20 msec later than the expected LV-RV conduction time) from the LV test pace. The first LV test pace (0.5V/0.9 msec) is ineffectual; loss of LV capture is indicated by emergence of native AV conduction (conduction time = 300 msec). The second LV test pace (1.0V/0.9 msec) is effective with LV-RV conduction time 180 msec, satisfying the capture detection window.

(Modified from Crossley GH, et al: Automated left ventricular capture management. Pacing Clin Electrophysiol 30:1190-1200, 2007.)

Atrioventricular conduction time is determined during overdrive (atrial pacing) simultaneous BiV pacing at a long pAVI sufficient to yield BiV sensing. This pAVI is set to 30 msec (nominal during LV-RV conduction check) + measured LV-RV conduction time + 30 msec. If no VS events during atrial pacing are seen during this time window, LVPTS can proceed. If any VS events occur within this time window, the test aborts. The difference between LV → RV and AV conduction time must therefore be consistently greater than about 60 msec to proceed with LV capture threshold determination. If the time difference between LV → RV and AV conduction is similar (e.g., <60 msec), loss of capture during LV monochamber pacing cannot be differentiated from native AV conduction. Anticipated increases in AVIs during overdrive atrial pacing increases the probability that AV conduction times will be sufficiently longer than LV-RV times. If AV conduction is absent (heart block), LVPTS proceeds. The AV conduction test is not performed in nontracking modes.

If these conditions are satisfied, LVPTS is initiated at the overdrive rate applied during VV and AV conduction tests (tracking modes). In nontracking modes (e.g., VVI/R) the overdrive rate on test cycles depends on the rate of the previous three “support” cycles. Biventricular pacing is initiated in a series of three or more support cycles at programmed ventricular voltage outputs, followed by single monochamber LV pacing “test” cycles (at pAVI 30 msec in dual-chamber modes), with progressively increasing LV output starting at 0.5 V per programmed pulse width below the last measured threshold. Consequently, a single ineffectual monochamber LV paced event resulting in a ventricular pause can occur in each test sequence, which explains the pacing “support” cycles. The LV capture detection window is defined as 30 msec earlier and 20 msec later than the longest observed LV-RV conduction time. LV capture is indicated by RV sense events within the capture window, whereas absent RV sensed events or events outside (either direction) the capture window of 30 msec or longer are interpreted as loss of LV capture. Loss of LV capture is defined as the output voltage in which two thirds of test stimuli fail. Reciprocally, LV capture is defined as the output voltage in which two thirds of test stimuli succeed. The LV capture amplitude threshold is the voltage at which three consecutive test cycles result in capture (see Fig. 31-21). If LVCM is programmed to Adaptive, the programmed Amplitude Safety Margin is added to the determined threshold with an upper limit defined by the Maximum Adapted Safety Margin parameter. If the new threshold has decreased with respect to the previously measured value, the LV amplitude is decreased by 1 programmable value. If the new threshold has increased with respect to the previously measured value, the LV amplitude is set to the measured threshold plus the LV amplitude safety margin. However, if the measured threshold plus the LV amplitude safety margin exceeds the LV maximum adapted amplitude, the LV amplitude is set to the LV maximum adapted amplitude. If a high-threshold condition occurs (e.g., capture is not determined at LV maximum adapted amplitude), LV amplitude is set to LV maximum adapted amplitude.

Notably, LVCM threshold determination is performed at 01:00 (1 am) (device clock). The reason for this is to minimize symptoms associated with intermittent loss of ventricular capture (single-cycle pauses), particularly in pacemaker-dependent patients.

Potential Limitations and Complications

A variety of clinical conditions must be satisfied for LVCM operation. Failure to differentiate AV versus VV conduction times is probably more likely to occur in patients with relatively normal AV conduction times, since VV conduction times are generally less than 100 msec. Overdrive atrial pacing reduces this possibility by stressing AV conduction. Extraneous sensed events in the AV and VV conduction detection windows (ventricular premature depolarization [VPDs], noise) may result in spurious LV threshold determinations. Intermittent loss and recovery of atrial capture during the AV conduction test can result in spurious LV threshold determinations. Consequently, another “test abort” criterion for uncertain atrial capture was added to the LCVM algorithm.

Single-cycle loss of ventricular capture during LVCM threshold determinations create ECG mimicry of pacing inhibition (crosstalk, conductor or insulation failure, EMI)⁹⁹ (Fig. 31-22). This may not be simply a cosmetic issue. Short-long-short sequences (“pauses”) terminated by single ventricular paced beats have been linked to ventricular tachycardia/fibrillation (VT/VF) onset and constitute a unique form of bradycardia pacing-induced ventricular proarrhythmia.¹⁰⁰ It seems likely that an example of VT/VF onset preceded by a short-long-short sequence during LVCM threshold determination will eventually be recorded.

Figure 31-22 Short-long-short sequences (pauses) and pseudo-crosstalk during LVCM operation.

Top left, Continuous AV sequential BiV pacing, followed by LV-RV conduction check and AV conduction check. The underlying rhythm was complete AV block, and therefore the AV conduction check was satisfied (absence of VS events within detection window). A sequence of pacing support cycles (indicated by increase in pacing rate) precedes the LV Pacing Threshold Search (LVPTS). The first LV test pace is ineffectual; in the absence of AV conduction and a ventricular escape mechanism, a pause occurs. This is followed by a pacing support cycle and another ineffectual LV test pace resulting in a second pause. The third LV test pace is effective, preventing a pause. Because of the short PAV (30 msec) and low-output voltage (0.5 V) during the LVPTS, only the atrial pacing stimulus is visible, mimicking true-crosstalk inhibition of ventricular output.

(Modified from Ho R, Mark GE, Rhim ES, Shorrock SM: Pseudo crosstalk behavior in a patient with atrio-ventricular block and implanted biventricular defibrillator. Pacing Clin Electrophysiol 30:1555-1557, 2007.)

Methods to Guarantee Ventricular Electromechanical Resynchronization

The translational mechanism of CRT is ventricular activation wavefront fusion.^12,72 Resynchronization of ventricular activation restores part or all of the LBBB-induced impairment in LV mechanics. The immediate consequence is improvement in contractility (pump function),^12,13,67,68 followed by reductions in global and regional structural abnormalities (reverse remodeling).^{83,92,93,101,102}

Therefore, the primary goal of CRT programming is maximal and continuous correction of LV conduction delay. Maximal means the greatest achievable correction of the delayed LV activation sequence and continuous means on every cardiac cycle for the patient’s lifetime. The following conditions must be specified to achieve this goal: (1) LV stimulation is delivered from a site capable of reversing LBBB conduction delay, (2) LV stimulation threshold is within the output capacity of the pulse generator, (3) LV pacing occurs on every cardiac cycle, and (4) LV timing is adjusted to guarantee ventricular activation fusion on every cardiac cycle. Conditions 1 and 2 are determined primarily by selection of a specific site for LV pacing and are largely fixed. Condition 3 is a dynamic process that requires continuous updating but is directly addressed by programming strategies and automaticity. Condition 4 is the intended end result of satisfying conditions 1 to 3 and provides evidence that maximal and continuous correction of LV conduction delay has been achieved.

Generating Ventricular Activation Wavefront Fusion during CRT

Analysis of ventricular activation using the standard 12-lead surface ECG accurately predicts the probability of LV reverse volumetric remodeling during CRT in patients with ventricular-asynchronous heart failure and LBBB.⁷² The probability of reverse remodeling is explained by interactions between substrate conditions and pacing-induced changes in ventricular activation before and after CRT. The main findings are that (1) the translational mechanism for volumetric remodeling is activation wavefront fusion, evident on the paced ECG, and (2) the probability of remodeling is positively influenced by LV conduction delay (maximum left ventricular activation time, LVAT_max) and negatively influenced by LV scar volume (QRS score) on the baseline ECG (Fig. 31-23). Longer LVAT and smaller LV scar volume during LBBB on the baseline ECG predict higher probability of reverse modeling (Fig. 31-24). Evidence for activation wavefront fusion on the CRT-paced ECG, indicating conduction defect reversal, also predicts higher probability of reverse remodeling (Fig. 31-25). The framework for analysis of ventricular activation using the 12-lead surface ECG has direct application to CRT programming.

Figure 31-23 Predicting reverse volumetric remodeling during CRT using analysis of ventricular activation on 12 lead surface ECG.

Baseline ventricular activation analysis during left bundle branch block (LBBB): longer maximum LV activation time (LVAT_max; LV conduction delay) predicts greater odds of reverse remodeling (30% relative increased odds for each 10-msec increase), whereas higher QRS scores (indicating % LV scar volume) predicts reduced odds of reverse remodeling (50% relative decreased odds for 1-point increase in QRS score of 1-4, 10% relative decreased odds for each 1-point increase above 4). Post-CRT analysis of ventricular activation using QRS hieroglyphs: Emergence of R waves in leads V1-V2 (indicating posterior → anterior activation reversal) predicts greater odds of reverse remodeling (2.76 relative increased odds for each 1-mV increase above 4.5 mV). Emergence of QR, Qr, or QS in lead I (indicating left → right activation reversal) resulting in frontal axis quadrant shift predicts greater odds of reverse remodeling (2.0 relative increased odds of reverse remodeling).

Figure 31-24 Probability of LV reverse remodeling by baseline and post-CRT predictors.

Top left, Increasing probability of response with greater maximum left ventricular activation time (LVAT_max). Line represents predicted probability of response for each value of LVAT_max. Diamonds represent actual response rates among LVAT_max quartiles. Top right, Increasing probability of response with greater LVAT_max and post-CRT evidence of ventricular fusion (P→A activation reversal). Bottom left, Increasing probability of response with greater LVAT_max and post-CRT evidence of ventricular fusion (L→R activation reversal). Bottom right, Increases in QRS score reduce response probability for any value of LVAT_max.

(From Sweeney MO, et al: Analysis of ventricular activation using surface electrocardiography to predict left ventricular reverse volumetric remodeling during cardiac resynchronization therapy. Circulation 121:626-634, 2010.)

Figure 31-25 Predicting LV reverse remodeling during CRT using a global measure of ventricular fusion.

Greater changes in R-wave amplitudes after CRT, indicative of wavefront fusion, predict higher probability of response. The points along the line are individual predicted values.

(From Sweeney MO, et al: Analysis of ventricular activation using surface electrocardiography to predict left ventricular reverse volumetric remodeling during cardiac resynchronization therapy. Circulation 121:626-634, 2010.)

Characterization of Ventricular Activation during LBBB Using QRS Hieroglyphic Analytic Framework Derived from Surface ECG

Typical LBBB activation is registered as R → L (frontal plane), A → P (horizontal plane), and variable axis. This yields a QRS hieroglyphic signature with dominant positive forces in I, aVL (R, Rs), negative forces in aVR (QS), variable forces in II, III, aVF (R, Rs, rS, QS), dominant negative forces in V1-V2 (QS, rS), transition V3-V5 (rS→Rs, R), and dominant positive forces in V5-V6 (R, Rs) (Fig. 31-26).

Figure 31-26 Characterization of ventricular activation during LBBB using QRS hieroglyphs.

Typical LBBB with dominant R in I (BIR glyph) and L (BLR glyph), indicating R → L activation; rS or QS in II, III, F, and rS or QS in V1-V3 (BV1QS, BV2QS glyphs), indicating anterior → posterior activation). In some patients, RS and Rs emerge in V4-V6. Left superior axis is caused by QS glyphs in leads II, III, aVF. LV activation plot (upper right) indicates R → L activation (red arrows, RV pacing wavefront; blue arrows, LV pacing wavefront).

QRS Notching

Left bundle branch block is characterized by sequential ventricular activation (RV → LV)^12,18,20 and registered as fragmented QRS complexes with RSR′ configuration (“notch”). QRS notching was defined as 1 notch or more in the R or S wave present in 2 or more adjacent anterior, lateral, or inferior leads¹⁰³ (Fig. 31-27).

Figure 31-27 Calculation of LV and RV activation time using surface ECG.

First downward arrow indicates QRS onset. Second downward arrow indicates first QRS notch, which represents transition between RV and LV activation. Time difference (msec) between first and second arrows is right ventricular activation time (RVAT). RVAT is measured in multiple anatomic regions determined by distribution of QRS notching. Shown are RVAT determinations for inferior and apical regions. LVAT_max is QRSd − RVAT_min = 160 msec − 45.3 msec = 114.7 msec.

Right and Left Ventricular Activation Time

Because multiple notches in the R and S wave during LBBB may result from myocardial scar,¹⁰³ the first notch was assumed to indicate the transition between RV and LV depolarization (see Fig. 31-27). Notching in the first 40 msec of the S wave in V1-V2 was excluded because this indicates scar in the QRS score. Right ventricular activation time (RVAT) was measured as time (msec) between QRS onset and first notch in any of two or more adjacent leads. LVAT was QRSd − RVAT (msec).

For modeling purposes, the longest LVAT (LVAT_max) recorded in any lead and region was used. A numerical relationship between QRSd and LVAT_max was derived using linear regression [LVAT_max (msec) = −35.839 + 0.763*QRSd (msec) + 0.000619*QRSd (msec)∧2], permitting estimation of LVAT_max in the absence of QRS notching (40 patients).

Quantification of Left Ventricular Scar

Myocardial scar has been linked to reduced CRT response¹⁰⁴ (Fig. 31-28). ECG quantification of LV scar volume was calculated using the Selvester QRS score for LBBB.^105,106 The effects of scar on LBBB surface registration translate as specific QRS hieroglyphic signatures, manifest as unopposed rightward electrical forces by infarct region: qR, QR, rS in I, aVL (anterior-superior); qR, QR, rS in I, V5-V6 (apical); qR, QR, rS in II, aVF (inferior); QS → rS, RS, or Rs in V1-V2 (anterior-septal). Notching on the ascending limb of the S wave in V1-V2 indicates posterolateral infarct.

Figure 31-28 Quantification of percentage LV scar volume using QRS score.

This patient has inferior (INF), anterosuperior (ASUP), anteroseptal (AS), posterolateral (PL), and apical (AP) scar, yielding QRS score = 14 (LV scar volume 32%).

Evidence for Ventricular Activation Fusion during CRT

Experimental models of LBBB demonstrate maximum improvement in LV pump function occurs when intra-LV electrical asynchrony is minimized by wavefront fusion¹² (see Figs. 31-29 to 31-44). Wavefront opposition and reversal during BiV pacing should yield ECG evidence of fusion: (1) expected change in frontal plane electrical axis (normal or LADEV→RADEV) and (2) expected changes in QRS hieroglyphic signatures. Rightward forces emerge in leads with dominant leftward forces (R in I, aVL → qR, QR, QS). Anterior forces emerge in leads with dominant posterior forces (QS in V1 → rS, RS, Rs, R; QS or rS in V2 → RS, Rs, R; rS or RS V3→ Rs or R, etc).

Figure 31-29 Ventricular activation during monochamber LV pacing from anterior coronary vein.

A, Approximately 75% of patients had LBBB activation sequence with minimal or absent PV1R during mid-distal anterior vein stimulation. B, Approximately 25% had right bundle branch block (RBBB) activation but only during very proximal anterior vein stimulation.

(Modified from Giudici MC, et al: Electrocardiographic patterns during pacing the great cardiac and middle cardiac veins. Pacing Clin Electrophysiol 30:1376-1380, 2007.)

Figure 31-30 Ventricular activation during middle cardiac vein pacing.

100% of patients had LBBB activation with left superior axis.

(Modified from Giudici MC, et al: Electrocardiographic patterns during pacing the great cardiac and middle cardiac veins. Pacing Clin Electrophysiol 30:1376-1380, 2007.)

Figure 31-31 Ventricular activation during lateral cardiac vein pacing (posterior wall).

100% of patients had RBBB activation with rightward axis.

(Modified from Giudici MC, et al: Electrocardiographic patterns during pacing the great cardiac and middle cardiac veins. Pacing Clin Electrophysiol 30:1376-1380, 2007.)

Figure 31-32 Evidence for ventricular activation fusion during CRT.

Same patient as in Figure 31-26. Note the following changes in global ventricular activation as compared to LBBB. R-wave emergence where there were previously S waves; Q, QS, or S wave emergence where there were previously R waves. R waves in leads I (BIR) and aVL (BLR) have been replaced by QS and qR complexes (PIQS and PLqR, respectively); indicating reversal of activation in the frontal plane (L→R). QS waves in V1 (BV1QS) and V2 (BV2QS) have been replaced by R waves (PV1R, PV2rS), indicating reversal of activation in the horizontal plane (P→A). LV activation plot (upper right) indicates fusion (red arrows, RV pacing wavefront; blue arrows, LV pacing wavefront).

Figure 31-33 Generation of ventricular activation fusion after CRT.

Side-by-side comparison global LV activation during LBBB (top) and post-CRT (bottom). Same patient as in Figures 31-26 and 31-30. Note reversal of activation by regions which yields a global pattern of ventricular activation fusion. This patient had QRS score = 1 (inferior scar), LVAT_max = 147 msec, and robust evidence of fusion post-CRT (left → right, posterior → anterior reversal). Predicted probability of ≥10% reduction in ESV at 6 months was 90%, actual ESV reduction was 24%.

(Modified from Sweeney MO, et al: Analysis of ventricular activation using surface electrocardiography to predict left ventricular reverse volumetric remodeling during cardiac resynchronization therapy. Circulation 121:626-634, 2010.)

Figure 31-34 Typical LBBB ventricular activation and corresponding LV longitudinal strain map.

LV conduction delay is registered as regional mechanical delay, indicated by early septal activation and delayed posterior wall activation.

Figure 31-35 Effect of “LV only” LVO pacing on ventricular activation (surface ECG) and LV strain map.

Same patient as in Figure 31-34. Baseline P-R interval is 240 msec. LVO pacing is delivered at atrioventricular delay (AVD) 160 msec. Note changes on the surface ECG indicative of incomplete ventricular fusion. QS in lead V1 has been replaced by rS; R in lead I has been replaced by qR. Frontal plane axis shift has not occurred. Corresponding strain maps demonstrate reduction in posterior wall delay; however, L→R, A→P activation persists.

Figure 31-36 Effect of LVO pacing on ventricular activation (surface ECG) and LV strain map.

Same patient as in Figures 34-34 and 34-35. Baseline P-R interval is 240 msec. LVO pacing is now delivered at AVD 120 msec. Note changes on the surface ECG indicative of ventricular activation fusion. QS in lead V1 has been replaced by RS; R in lead I has been replaced by QS. Right superior axis has replaced normal axis. Corresponding strain maps demonstrate synchronization of septal and posterior wall activation. LV volume is visually reduced.

Figure 31-37 Effect of LVO pacing on ventricular activation (surface ECG) and LV strain map.

Same patient as in Figures 34-34 to 34-36. Baseline P-R interval is 240 msec. LVO pacing is now delivered at AVD 100 msec. Note changes on the surface ECG indicative of complete reversal of ventricular activation: QS persists in lead I, qR in lead AVL has been replaced by qr (R regression), and QS in leads V1-V3 have been replaced by Rs. Corresponding strain map demonstrates early activation of posterior wall and delayed activation of septum. Activation is now R → L, P → A.

Figure 31-38 Lead positions and baseline LBBB activation.

Same patient as in Figure 39-41. Top, RVA and LV posterior wall (lateral coronary vein) stimulation sites. Bottom, Typical R → L (frontal), A → P (horizontal) activation sequence during LBBB.

Figure 31-39 QRS hieroglyphic signatures during monochamber RV and LV pacing.

Top, Monochamber RVA pacing activation is typical. Bottom, Monochamber LV pacing from posterior wall (lateral coronary vein) demonstrates typical dominant R leads V1-V3, indicating P→A activation. Atypical dominant qR in lead I, aVL is caused by “lead I paradox.”

Figure 31-40 Incomplete ventricular activation fusion during simultaneous biventricular pacing.

Top, Intrinsic AVI (P-R interval surrogate) is 160 msec, AVD is 100 msec. Monochamber RV paced → LV sensed = 120 msec; monochamber LV paced → = 120 msec (equivalent sequential ventricular conduction times between pacing sites). Global activation closely resembles monochamber RV activation (see Fig. 31-39, top). Incomplete fusion is indicated by slight QRS narrowing, most evident in leads V1-V3. Overall activation pattern remains R→L, A→P. Bottom, AVD remains fixed at 100 msec. Persistent incomplete fusion despite sequential BiV stimulation with LV 80 msec preceding RV. Minimal change in activation sequence compared to simultaneous BiV pacing. Note slight attenuation of lead I and aVL R, emergence of small monophasic R in V4-V6, and further shortening of QRS duration in V1-V2.

Figure 31-41 Maximal evidence of ventricular activation fusion during sequential biventricular (BiV) pacing at shortened atrioventricular delay (AVD).

Sequential BiV pacing with LV 80 msec preceding RV. AVD shortened from 100 msec → 80 msec. Evidence of activation reversal in pivotal leads is now present. Note leads I and aVL R → qR, leads V1-V2 QS → RS, lead V3 rQr′ → R. Since LV is always delayed relative to RV during LBBB, shortening AVD advances LV further than maximally achievable with sequential ventricular pacing at longer AVD.

Figure 31-42 Lead positions and baseline LBBB activation.

Same patient as in Figures 31-43 and 31-44. Top, Radiographs of RVA and LV posterior wall (lateral coronary vein) stimulation sites. Bottom, Typical R → L (frontal), A → P (horizontal) activation sequence during LBBB.

Figure 31-43 Atypical QRS hieroglyphic signatures during monochamber RV, LV, and BiV pacing.

Top, Monochamber right ventricular apex (RVA) pacing activation. Note unexpected lead I rS, V1 R, and V2 Rs. Frontal plane axis is right superior. This activation pattern is likely explained by conduction blocks caused by extensive myocardial scar (coronary artery disease). Bottom, Monochamber left ventricular (LV) pacing from posterior wall (lateral coronary vein). Note atypical lead I RS, whereas lead aVL QR and lead V1-V5 R are typical. The predominantly positive force in lead I during LV posterior wall stimulation may be caused by scar. This can also occur without scar (lead I paradox), however, most likely explained by a basal location of the LV lead and more horizontal anatomic axis of the LV or posterior rotation of the heart. Consequently, LV activation proceeds paradoxically from R → L, which yields a dominant R in lead I.

Figure 31-44 Generating evidence for ventricular activation fusion during BiV pacing for atypical monochamber RV and LV activation patterns.

Top, Simultaneous biventricular (BiV) pacing. Global activation closely resembles monochamber right ventricular (RV) activation (see Fig. 31-43, top). Incomplete fusion is indicated by slight attenuation of the dominant component of lead I rS, lead aVL R, and slight QRS narrowing, most evident in leads V1-V2. Note that dominant R in V1-V2 (present during monochamber RV and LV pacing) persists. Bottom, Sequential BiV pacing with LV 60 msec preceding RV. Activation directionality is reversed in multiple leads. Note BIQS → PI qRs, BaVLR → PAVL qR, BII, aVF, III QS → PII, aVF, III rS. Note emergence of Rs in V3-V5.

Monochamber LV pacing generates specific ventricular activation patterns generally in accordance with stimulation site (see Figs. 31-29 to 31-31). Anterior vein stimulation most often results in LBBB activation sequence; infrequently, very proximal sites in the anterior vein yield RBBB activation. Middle cardiac vein stimulation always results in LBBB activation sequence, duplicating RVA pacing. Lateral cardiac vein (posterior wall) typically results in complete reversal of the LBBB activation sequence (e.g., RBBB, rightward axis).

However, nearly identical ECG activation sequences may be registered at widely spaced sites,¹⁰⁷ and electrical resynchronization during CRT correlates unpredictably with stimulation site because of conduction blocks.²⁰ For these reasons, knowledge of specific LV stimulation site is not relevant to this approach for generating evidence of ventricular activation fusion, which emphasizes changes in ventricular activation indicated by analysis of QRS hieroglyphic patterns.

CRT Programming to Generate Maximum Evidence of Ventricular Activation Fusion

As discussed previously, optimal ventricular resynchronization occurs when the wavefronts from the RV pacing site (or originating from RBBc) and the LV pacing site collide halfway⁷⁸ (see Fig. 31-10). Experimental and human evidence indicates that optimal hemodynamic effect can be achieved using LV pacing at certain atrioventricular delays (AVDs) as well as using simultaneous and sequential BiV pacing, because fusion of electrical wavefronts can be achieved by either approach^13,79 (see Figs. 31-32 and 31-33; see also Fig. 31-11).

The interaction among AVD, VV timing, and generation of activation fusion is most easily understood during “LV only” (LVO) pacing (see Fig. 31-34). In this situation, ventricular activation is controlled by fusion between LV pacing and RBBB conduction. The P-R interval is a surrogate for left-sided AV timing. Any AVD shorter than the baseline P-R interval will advance LV activation. Partial correction of LV conduction delay at a relatively long AVD < PR is registered on the surface ECG as incomplete ventricular fusion and delayed time to peak strain on the posterior wall relative to septum (see Fig. 31-35). Complete correction of LV conduction delay at shorter AVD is registered as ventricular activation fusion and synchronization of time to peak strain between septum and posterior wall (see Fig. 31-36). Overcorrection of LV conduction delay (complete reversal of activation) at even shorter AVD is registered as L → R, P → A activation on the surface ECG and delayed time to peak strain on the septum relative to posterior wall (see Fig. 31-37).

More often, CRT is delivered with biventricular pacing. In this situation, evidence for generation of ventricular activation fusion on the surface ECG consists of two pacing wavefronts with unique ECG signatures. Certainty regarding maximal ventricular activation fusion during BiV pacing therefore requires characterization of ventricular activation during monochamber RV and LV pacing (see Figs. 31-19 and 31-20). This is most accurately accomplished in the ventricular-only pacing mode at a rate exceeding the prevailing ventricular rate, so as to eliminate the possibility of fusion with native activation. Once the ECG signatures of monochamber RV and LV pacing have been characterized, simultaneous BiV pacing is initiated at an AVD sufficiently short to preempt native ventricular activation (e.g., 50% of baseline P-R interval). This allows complete control of ventricular activation timing by replacing intrinsic activation with paced activation. Since LV activation is always delayed relative to RV activation during LBBB, and the pacemaker AVD is recorded between the right atrium and right ventricle, shortening the AVD advances the LV activation and titrates the dose of resynchronization. Further advancement of LV activation is possible with sequential ventricular stimulation. The ECG is then examined for evidence of activation fusion using a similar QRS analytic architecture as previously described.

Occasionally, maximal attainable sequential ventricular stimulation does not achieve ventricular activation fusion, even when the AVD is substantially shorter than the P-R interval. In this situation, further shortening of the AVD is necessary to advance the LV activation sufficiently and achieve maximal evidence of ventricular activation fusion (see Figs. 31-38 to 31-41). In this manner, the AV delay and V-V interval display a necessary and logical interaction as control parameters for ventricular activation timing.

Generally, RVA pacing produces a ventricular activation sequence that duplicates the QRS hieroglyphic signature of LBBB. Under these conditions, typical changes in frontal plane axis and QRS hieroglyphs indicative of ventricular activation fusion are observed (see Figs. 31-29 to 31-32). Occasionally, atypical ventricular activation patterns occur during monochamber RVA and LV pacing and confound the typical QRS hieroglyphic signature for ventricular activation fusion during BiV pacing (see Figs. 31-42 to 31-44). In this situation, typical QRS glyph changes in the expected direction during BiV pacing are not observed. Evidence for ventricular activation fusion must therefore rely on the principle of “activation wavefront reversal” in pivotal leads. The use of a global measure of activation fusion is particularly useful in this situation (see earlier) because this eliminates directionality, which can be confounded by atypical pacing activation patterns, in favor of magnitude of change opposite the initial forces in pivotal ECG leads.

Methods to Guarantee Optimal Atrioventricular Resynchronization

Correction of ventricular conduction delay to improve ventricular mechanics and induce reverse volumetric remodeling is the primary target of CRT. Resynchronization of ventricular electromechanical timing directly influences systolic performance. Since the timing of ventricular stimulation is primarily controlled with the pacemaker A-V interval, CRT also has direct effects on AV timing, which influences ventricular loading conditions and diastolic function. Diastolic dysfunction is common in systolic heart failure and contributes to symptoms. Optimal AV resynchronization can reduce diastolic dysfunction and improve symptoms. The interaction among the pAVI, ventricular electromechanical resynchronization, and AV resynchronization has resulted in much confusion and debate regarding CRT programming considerations. This confusion has two sources: (1) failure to understand the interaction between pAVI, AV timing, and VV timing (see earlier) and (2) a long-standing historical interest in improving pump function by optimizing diastolic function during conventional dual-chamber, monoventricular (RV) pacing. In the latter situation, the effects of changes in pAVI were not confounded by simultaneous effects on ventricular resynchronization and therefore easier to understand.

Some generalizations regarding the pacemaker AVI, ventricular resynchronization, and AV resynchronization during CRT are useful, as follows:

Figure 31-45 Different effects of change in loading conditions in normal and failing ventricles.

Left, The failing ventricle operates on a depressed Frank-Starling relationship where smaller increases in stroke volume are observed for any increase in preload compared to the normal ventricle. Right, At the same time, larger increases in left ventricular (LV) diastolic pressures are observed for any increase in diastolic volume for failing ventricles compared to normal ventricles. This diastolic dysfunction results in increased LV diastolic volumes from ventricular enlargement (remodeling), causing higher LV pressures, which do not augment contractility but worsen symptoms.

(Modified from Marcus RH, Harry MJ, Lang RM: Pacemaker therapy in refractory heart failure: a systematic approach to electromechanical optimization of diastolic function. JACC Cardiovasc Imaging 2:908-913, 2009.)

Consequently, persistent diastolic dysfunction, despite improvement in LV size and contractility, is an important and frequently unrecognized source of symptoms in asynchronous heart failure patients treated with CRT.

Effects of Pacemaker Atrioventricular Interval on Loading Conditions, Diastolic Function, and Systolic Function

The events of the cardiac mechanical cycle are again displayed in Figure 31-46. Ventricular filling occurs during diastole and has two components: early passive filling (E wave) and later active filling (A wave). The velocity of the diastolic filling waves is determined by the LA-LV transmitral pressure (TMP) gradient, which is phasic. The timing of the E wave is controlled by the preceding ventricular contraction. The timing of the A-wave is fixed by the sinus rate (or atrial pacing rate). The timing of the E wave relative to the A wave is influenced by the preceding ventricular contraction and the AV conduction time (e.g., P-R interval).

Figure 31-46 Left ventricular diastolic event timing relationships.

Forward flow in diastole is composed of early passive LV filling (E wave) and late active LV filling (A wave). The E wave is always timed to the prior ventricular contraction, whereas the timing of the A wave is fixed to atrial systole. Isovolumetric contraction time is the interval between mitral valve closure and aortic valve opening. Isovolumetric relaxation time is the interval between aortic valve closure and mitral valve opening.

Diastolic function is impaired in asynchronous heart failure. The left ventricle operates at higher diastolic pressures with the consequence that filling times are shortened (Fig. 31-47). Ventricular conduction delay and reduced pump function result in prolonged isovolumetric contraction and relaxation times, which shorten diastole (Fig. 31-48). These derangements worsen at the higher heart rates typically observed in heart failure patients.

Figure 31-47 Diastolic events and pressure gradients under different loading conditions.

Pressure-time curves from an anesthesized dog are shown. The first LA-LV crossover indicates the end of isovolumetric relaxation (IR), mitral valve opening, and early rapid filling when LA pressure exceeds LV pressure (E wave on the mitral inflow velocity). The solid arrow indicates minimum LV pressure. Peak mitral E-wave velocity corresponds to the second pressure crossover; LV pressure now exceeds LA pressure, decelerating mitral inflow, and concluding the early rapid filling phase. The slow filling phase is characterized by equalization of pressures and minimal LV filling. During atrial contraction, LA pressure exceeds LV pressure and mitral flow increases (active filling phase). The dotted arrow indicates the LV pressure preatrial contraction; the dashed arrow indicates LV end-diastolic pressure (EDP). The upper tracing was obtained at normal LVEDP (~8 mm Hg), the lower panel at high LVEDP (~24 mm Hg). Pressure differences are larger because of reduced compliance; LA contraction barely exceeds LVEDP, reducing active LV filling.

(Modified from Nagueh et al: Recommendations for the evaluation of left ventricular diastolic function by echocardiography. J Am Soc Echocardiogr 22:107-133, 2008.)

Figure 31-48 Effect of increased contraction/relaxation times on diastolic function in failing ventricle.

IVCT, Isovolumetric contraction time; IVRT, isovolumetric relaxation time; A, mitral regurgitation flow duration; B, left ventricular (LV) outflow duration; C, LV inflow duration (diastole). Index of diastolic performance = A − B/C.

(Modified from Marcus RH, Harry MJ, Lang RM: Pacemaker therapy in refractory heart failure: a systematic approach to electromechanical optimization of diastolic function. JACC Cardiovasc Imaging 2:908-913, 2009.)

On the standard echocardiogram, LV inflow velocities are used to characterize diastolic function. Characteristic changes in mitral inflow are observed in response to changes in the TMP gradient (Fig. 31-49). The mitral E-wave velocity is determined by the early-diastolic LA-LV pressure gradient and reflects preload and alterations in LV relaxation. The mitral A-wave velocity is determined by late-diastolic LA-LV pressure gradient and reflects LV compliance (pressure-volume relationship) and LA contractile function. E-wave deceleration time reflects LV relaxation, LV diastolic pressures, and LV compliance.

Figure 31-49 Changes in LV inflow velocity patterns in response to changes in transmitral pressure gradient.

Left, Normal LA-LV pressure relationships and corresponding LV inflow velocity pattern. Note high E-wave velocity with rapid deceleration, E/A > 1. Middle, Impaired relaxation results in higher LV diastolic pressures, which reduce E-velocity (E/A < 1) and increase E-wave deceleration time. Note, a small E-velocity could also be caused by low left atrial pressure (LAP), which must be excluded by independent assessment of LA pressure. Right, Compensatory increase in LA pressure to maintain LV filling results in “pseudonormalization,” indicated by high E-wave velocity (E/A >1) and very rapid E-deceleration time.

(Modified from Nagueh, et al: Recommendations for the evaluation of left ventricular diastolic function by echocardiography. J Am Soc Echocardiogr 22:107-133, 2008.)

Changes in preload modify the echocardiographic patterns of diastolic function (Figs. 31-50 and 31-51). In the normal pattern, the majority of forward flow occurs in early diastole, yielding E/A ratio greater than 1.5 and short E-wave deceleration time (<180 msec) and isovolumetric relaxation time (<80 msec). Incomplete relaxation caused by myocardial disease results in higher early-diastolic LV pressures and reduced TMP gradient. Consequently, E-wave velocity and E/A ratio decrease (<1), and isovolumetric relaxation time (>90 msec) and E-wave deceleration time (>220 msec) increase (stage I diastolic dysfunction, relaxation abnormality). The compensatory response to maintain LV filling is an increase in LA pressure, resulting in higher E-wave velocity and E/A ratio. Stage II diastolic dysfunction is characterized by E/A more than 1, isovolumetric relaxation time less than 90 msec, and deceleration time less than 220 msec, termed pseudonormalization. Progression of diastolic dysfunction is indicated by a restrictive filling pattern, characterized by E/A over 1, isovolumetric relaxation time under 70 msec, and deceleration time under 150 msec, termed restrictive (stage III). A restrictive LV inflow pattern unresponsive to preload reduction is termed irreversible (stage IV). Stages II to IV are associated with increased LV filling pressures.

Figure 31-50 Echocardiographic filling patterns of diastolic dysfunction.

See text for details.

(Modified from Swedberg K, et al: Guidelines for the diagnosis and treatment of chronic heart failure: executive summary (update 2005). Eur Heart J 26(11):1115-1140, 2005.)

Figure 31-51 Echocardiographic patterns of diastolic function before and after preload reduction.

A, LV inflow velocity patterns as described in Figure 31-49. Pulmonary venous flow is determined by the pressure gradient between the pulmonary veins and the left atrium. Systolic forward flow (S) occurs during LA relaxation, diastolic forward flow (D) follows early LV filling. During LA contraction in late diastole, flow reversal into the pulmonary veins occurs (not shown). Increased early diastolic filling pressure results in decreased systolic flow and increased diastolic flow in the pulmonary veins. B, Response of diastolic filling patterns to preload reduction (Valsalva maneuver). Reduced preload results in reduced E-wave velocity and prolongation of E-deceleration time, whereas A-wave velocity is unchanged or increases. A ≥50% decrease in E/A ratio with Valsalva maneuver is highly specific for increased LV filling pressures. Failure of the LV inflow pattern to change with preload reduction is indicative of severe diastolic dysfunction (stage IV, irreversible).

(Modified from Marcus RH, Harry MJ, Lang RM: Pacemaker therapy in refractory heart failure: a systematic approach to electromechanical optimization of diastolic function. JACC Cardiovasc Imaging 2:908-913, 2009.)

The response of these LV inflow velocity patterns to preload reduction can be used to understand the relationship between diastolic function and changes in pacing rate and pAVI. Higher pacing rates combined with delayed isovolumetric relaxation time result in fusion of early and late LV filling, simulating stage III (restrictive) diastolic dysfunction (Fig. 31-52). Reduced pacing rates delay atrial contraction, displace late LV filling relative to delayed early filling, and increase diastolic filling time.

Figure 31-52 Effect of change in pacing rate on diastolic function.

Left, Pacing rate 70 pulses per minute (ppm). Prolonged IVRT results in delayed early filling resulting in E-A fusion. This simulates a stage III (“restrictive”) filling diastolic pattern. Right, Pacing rate 60 ppm. E wave is tied to prior ventricular contraction; A wave is now displaced rightward in time; consequently, diastolic filling periods increase from 200 → 400 msec, and E-A separation emerges (stage III → stage I diastolic dysfunction). Note index of diastolic function decreased from 1.45 → 0.73 (reflecting increase in C term in equation in Fig. 31-48).

(Modified from Marcus RH, Harry MJ, Lang RM: Pacemaker therapy in refractory heart failure: a systematic approach to electromechanical optimization of diastolic function. JACC Cardiovasc Imaging 2:908-913, 2009.)

Inappropriately short pAVIs result in truncation of late-diastolic active LV filling, indicated by minimization of the A-wave velocity, simulating stage III diastolic dysfunction. In extreme cases, atrial transport block and pulmonary flow reversal can occur. Increases in pAVI increase diastolic filling time, allowing for emergence of late-diastolic active LV filling (A-wave velocity emergence) and transformation to stage II diastolic dysfunction in this example (Fig. 31-53). In extreme cases, atrial transport block and pulmonary flow reversal can occur, triggering counterphysiologic neurohormonal responses (Fig. 31-54).

Figure 31-53 Changes in LV inflow velocity patterns in response to changes in transmitral pressure gradient during manipulation of the pacemaker AVI.

Left, Pacemaker atrioventricular interval (AVI) of 100 msec results in truncation of late-diastolic active filling (A wave is abolished). Progressive increase of pacemaker AVI to 150 msec (middle) and 200 msec (right) restores late-diastolic active filling (A-wave emergence).

(Modified from Marcus RH, Harry MJ, Lang RM: Pacemaker therapy in refractory heart failure: a systematic approach to electromechanical optimization of diastolic function. JACC Cardiovasc Imaging 2:908-913, 2009.)

Figure 31-54 Atrial transport block caused by inappropriately short pacemaker AVI.

Left and middle, Left ventricular (LV) inflow velocities and corresponding pulmonary vein velocities at long and short atrioventricular interval (AVI). Right, Catheter measurements of left-sided heart pressures at various representative pacemaker AVIs (short, optimal, long) in an animal model.

Pacemaker AVI 100ms results in atrial contraction after the onset of ventricular systole (note sudden deceleration in A-velocity, top left panel). LV contraction against a closed mitral valve results in marked increase in atrial pressure (downward arrow bottom left panel and first downward arrow in right panel where LA pressures are shown in green) and reversal of pulmonary venous flow (downward arrow bottom left panel and first upward arrow right-hand panel where pulmonary venous pressures shown in blue). Increase in pacemaker AVI to 200 msec allows emergence of late active filling (A wave) before onset of ventricular systole (note slower deceleration of A wave, top middle panel) and restoration of pulmonary venous forward flow (bottom middle panel, right-hand panel where pulmonary venous pressures are shown in blue). Diastolic dysfunction pattern changes from stage II → stage I when pacemaker AVI 100 → 200 msec. Note, long pacemaker AVI 300 msec duplicates counterphysiology of short AVI 100 msec (increase in LA pressure, pulmonary venous flow reversal caused by atrial contraction coincident with prior ventricular contraction).

(Left and middle panels modified from Marcus RH, Harry MJ, Lang RM: Pacemaker therapy in refractory heart failure: a systematic approach to electromechanical optimization of diastolic function. JACC Cardiovasc Imaging 2:908-913, 2009; right panel courtesy Douglas Hettrick, PhD, Minneapolis.)

Inappropriate long pAVIs result in collision of early and late diastolic filling (Figs. 31-54 and 31-55). The physiologic consequences are identical to first-degree AV block. Early passive diastolic filling collides with late active filling because the atrial contraction is displaced leftward in time relative to the next ventricular paced beat. Increase in left atrial (LA) pressure with atrial contraction interrupts pulmonary vein diastolic forward flow; late-diastolic LV → LA pressure gradient results in diastolic mitral regurgitation. In extreme cases, atrial contraction occurs during the preceding ventricular contraction, resulting in atrial transport block. This duplicates the situation of inappropriate short pAVI. A shorter pAVI moves early LV filling leftward (earlier) in time, since the E wave is “tied” to the prior ventricular contraction. This has the effect of separating the E wave and the oncoming A wave, which is fixed to atrial systole.

Figure 31-55 Atrioventricular decoupling caused by excessively long pacemaker AVI.

Left, Pacemaker atrioventricular interval (AVI) 250 msec results in fusion of early and late diastolic filling (E-A fusion). Atrial contraction during early LV filling interrupts pulmonary vein diastolic forward flow (downward arrow). Diastolic mitral regurgitation results from late diastolic LV → LA pressure gradient. Right, Pacemaker AVI 120 msec results in emergence of late diastolic active filling (A-wave velocity emergence), normal pulmonary vein diastolic forward flow, and elimination of diastolic mitral regurgitation.

(Modified from Marcus RH, Harry MJ, Lang RM: Pacemaker therapy in refractory heart failure: a systematic approach to electromechanical optimization of diastolic function. JACC Cardiovasc Imaging 2:908-913, 2009.)

Optimization of AV Resynchronization

Methods for titrating optimal AV resynchronization during multisite pacing therapy include echocardiographic analysis of LV inflow velocity patterns, EGM timing analysis derived from invasive hemodynamic monitoring, and real-time hemodynamic monitoring during manipulation of the pAVI. The common goal of these methods is to identify the single pAVI that yields (1) maximum improvement in LV preload and (2) maximum diastolic filling time. This requires that the timing of LA contraction to LV contraction be optimized. When LA contraction occurs too early relative to LV contraction, late diastolic active filling (A wave) fuses with early diastolic passive filling (E wave), reducing LV preload. This form of diastolic dysfunction occurs spontaneously during first-degree AV block, frequently accompanies LBBB, and is corrected by the short pAVIs necessary to achieve maximum evidence of ventricular fusion activation. An insufficiently short pAVI, particularly in the presence of significant first-degree AV block (long P-R interval) could also result in this form of diastolic dysfunction. This is commonly referred to as a pAVI that is “too long” relative to the intrinsic A-V interval (iAVI). A different form of diastolic dysfunction is induced when LA contraction occurs simultaneously, or immediately after LV contraction. Late diastolic filling (A-wave) is truncated by mitral valve closure and in the most extreme form results in atrial transport block. This reduces LV preload; causes increased LA pressure and pulmonary venous flow reversal, yielding a pattern of restrictive diastolic dysfunction; and can trigger counterphysiologic neurohormonal reflexes. This derangement never occurs spontaneously. However, these timing abnormalities can be inadvertently induced by a pAVI that is “too short” relative to the iAVI. Since early passive filling (E wave) is tied to the prior ventricular contraction, and late active filling (A wave) is independently timed by the sinus rate, the pAVI controls the timing relationship between A wave and ventricular contraction.

These considerations frame the dominant clinical concern and challenge to CRT programming regarding AVI management. Because generation of maximum evidence of ventricular activation wavefront fusion requires that BiV paced activation replace native ventricular activation, pAVIs substantially shorter than the iAVI are applied, usually accompanied by sequential ventricular timing (e.g., LV stimulation first, see later). Consequently, it is much less likely that pAVIs during CRT will be “too long.” It is much more likely that pAVIs sufficiently short to generate maximum evidence of ventricular activation fusion will inadvertently induce truncation or block of late-diastolic active (A-wave) filling (e.g., pAVI “too short”). This is a particular concern when the baseline P-R interval (iAVI) is short (e.g., <200 msec).

Conventional Echocardiography

Left Ventricular Inflow Analysis

Several methods to assess and tailor AV resynchronization using analysis of LV inflow velocities have been described. These methods involve real-time iterative manipulation of the pacemaker AVI and visual analysis of LV inflow patterns. The endpoint is the pAVI that times mitral valve closure coincident with the end of the untruncated A wave. The assumption is that the pAVI generated by this approach will (1) maximize LV preload (Frank-Starling) and (2) maximize diastolic filling time.

According to the method of Ritter et al.,¹⁰⁸ optimal pacemaker AVI during atrial sensing (SAVI) can be stated algebraically as: SAVI_optimal = SAVI_short + d, where d = (SAVI_long + QA_long) − (SAVI_short + QAI_short), Q = ventricular pacing stimulus, and A = termination of A wave.¹⁰⁹ This process is performed in three steps, as follows:

1 SAVI_long and QA_long are determined by programming a “long” sensed AVI (SAVI_long). SAVI_long is the longest AVI that results in (a) ventricular capture without fusion of native conduction and (b) spontaneous closure of the mitral valve before LV ejection (Fig. 31-56). QA_long is then measured as the time from the ventricular pacing stimulus to the end of the A wave.

2 SAVI_short and QA_short are determined by programming a “short” sensed AVI (SAVI_short). SAVI_short is the longest AVI that results in A-wave truncation (Fig. 31-57). QA_short is then measured as the time from the ventricular pacing stimulus to the end of the A wave.

3 The SAVI_optimal is then calculated from these values. AV optimization is confirmed by noting the return of normal E- and A-wave separation indicating improved diastolic filling time and optimized AV timing relationship (Figs. 31-58 and 31-59).

Figure 31-56 Ritter method for atrioventricular resynchronization: step 1.

Determination of optimal pacemaker atrioventricular interval during atrial sensing: SAVI_long and QA_long. AVI 160 msec is the longest AVI that yields full ventricular capture without A-wave truncation. The time from the ventricular pacing stimulus to the end of the A wave is QA_long.

Figure 31-57 Ritter method for AV resynchronization: step 2.

Determination of SAVI_short and QA_short. The SAVI is shortened until A-wave truncation is observed. This is SAVI_short. The time from the ventricular pacing stimulus to the end of the A wave is QA_short.

Figure 31-58 Ritter method for AV resynchronization: step 3.

Calculation of SAVI_optimal. Note restoration of normal E-A wave separation.

Figure 31-59 Ritter method for AV resynchronization: summary 1.

Left: SAVI_short = 60 msec and QA_short = 100 msec. Right: SAVI_long = 160 msec and QA_long = 20 msec. Optimal AVI = 60 + [(160 + 20) − (60 + 100)] = 60 + 20 = 80 msec.

Figure 31-60 summarizes an example of this approach.

Figure 31-60 Ritter method for AV resynchronization: summary 2.

Top, LV inflow velocity pattern during native AV conduction. Note E-A wave fusion from marked first-degree AV block. Bottom, Change in LV inflow velocity pattern across a range of SAVIs. SAVI = 60 msec results in A-wave truncation (too short). SAVI = 140 msec results in E-A wave fusion. SAVI = 80 msec results in E-A separation without A-wave truncation. Note diastolic filling pattern is now stage I (normal).

Meluzin et al.¹¹⁰ described a “simplified” two-step approach to the Ritter method. First, a “long” sensed AVI that results in (a) ventricular capture without fusion of native conduction and (b) spontaneous closure of the mitral valve before LV ejection is selected. The time between the end of the A wave (representing the end of the diastolic filling period) to the time of onset of high-velocity systolic mitral regurgitation (representing the onset of ventricular contraction) is recorded. Second, the time from the end of the A wave to the onset of low-velocity diastolic mitral regurgitation is denoted as t1. The optimal AVI is then calculated as the “long” AVI-t1. This approach accurately predicted the optimal AVI based on simultaneous invasive hemodynamic measurements in 78% of patients. Similar limitations as the Ritter method apply; also, at least mild mitral regurgitation is required for timing measurements.

The Ritter method is a tedious process. The value of quantification using the Ritter “formula” is unclear. A similar result could be achieved by simply manipulating the pAVI across a range of values while examining the LV inflow velocity pattern until E-A wave separation unaccompanied by A-wave truncation is observed (see Fig. 31-58). Although many published clinical investigations cite the “Ritter method” for AV resynchronization, only visual analysis of the LV inflow pattern was likely performed (e.g., algebraic calculations were omitted). As with all methods for AV synchronization using visual analysis of LV inflow patterns, subjective error is common because the terminal portion of the A wave is often difficult to discern. The Ritter method was developed for AV resynchronization in complete AV block and during monochamber RV pacing. Ventricular conduction is entirely different during biventricular pacing.

However, a potentially more serious problem with the Ritter and related methods is the fundamental assumption that timing ventricular contraction to the terminal portion of the A wave optimizes preload; this appears to be physiologically incorrect in the setting of heart failure with low ejection fraction. Insights regarding optimal AV resynchronization in the failing ventricle are generated by invasive hemodynamic studies that use the acute response of aortic pulse pressure, or LV +dP/dt_max, which are useful indices because they correlate with stroke volume and global contractile function. These studies indicate that the maximum acute improvement in LV pump function occurs when the left-sided atrioventricular mechanical latency (AVL) time is closest to zero (Fig. 31-61). This corresponds to the time of peak velocity of the A wave. Timing the onset of LV contraction to the end of LA emptying (e.g., A-wave termination, Ritter method) results in delayed AVL time and reduced hemodynamic response. As mentioned previously, the Ritter method was developed in pacemaker patients with normal LV function and AV block, a situation where LV pressure and volumes are low and compliance is high. However, in the failing ventricle, LV pressures and volumes are high and compliance is low (see Fig. 31-45). Consequently, onset of LV contraction at the highest point of LA pressure generation may be necessary for maximum hemodynamic improvement.

Figure 31-61 Optimal AV resynchronization using invasive hemodynamic monitoring.

Left graph, Example of LV pressure response during BiV pacing (a) and intrinsic ventricular conduction (b), accompanied by intrinsic LV electrogram (c) and intrinsic RA electrogram (d). A_P indicates the small presystolic peak in LV pressure caused by atrial contraction, and L_S indicates the onset of LV pressure development during isovolumetric contraction, taken as the first point with slope greater than 10% of maximum rate of increase of LV pressure. The time from the RA EGM to A_P is the interatrial electromechanical delay. In the absence of pacing, the RA-L_S time indicates the electromechanical AV delay. The interval (A_PL_S) between A_P and L_S is the atrioventricular mechanical latency (AVL). AVL reflects the time between the end of LA contraction and the onset of LV contraction. Therefore, a positive AVL (LA precedes LV) indicates LA contraction contributes to LV filling, whereas a negative AVL (e.g., pAVI “too short”) indicates LV contraction precedes LA emptying (truncation, transport block). During BV pacing, the LV electromechanical activation is advanced and the L_S point moves from right → left, indicated by a shift in LV pressure curve from a → b. The LV pressure curves are time-aligned against RA activation. Right panel, Acute hemodynamic changes in relation to different AVL values during manipulation of the pacemaker AVI. A and B, Responders (≥5% increase in pulse pressure) had maximum hemodynamic improvement at pAVI of about 40% to 50% of the intrinsic A-V interval (iAVI). Further shortening of the pAVI reduced hemodynamic response. For responders, the maximum hemodynamic improvement occurred when AVL ~ 0 msec, which occurs at a pAVI that did not decrease LVEDP. Hemodynamic response was reduced at positive AVL values (pAVI “too long”) and negative AVL values (pAVI “too short”). C and D, Nonresponders (<5% increase in pulse pressure) had no improvement at any pAVI and were made worse as the pAVI shortened.

(From Auricchio A, et al: Long-term clinical effect of hemodynamically optimized cardiac resynchronization therapy in patients with heart failure and ventricular conduction delay. J Am Coll Cardiol 39:2026-2033, 2002.)

With techniques for AV optimization using invasive LV pressure monitoring, some evidence indicates that acute hemodynamic response to CRT predicts better outcomes.^68,77,111 However, pulse pressure and LV +dP/dt can be confounded by changes in preload and afterload. Although this technique is useful for assessing the acute effects of manipulating critical pacing control parameters, ventricular stimulation sites, and ventricular sequencing on LV pump function, it is impractical for routine clinical care.

Left Ventricular Outflow Analysis

Measurement of the continuous-wave Doppler aortic velocity time integral (VTI) provides an estimate of stroke volume (Figs. 31-62 and 31-63). Changes in loading conditions during manipulation of critical pacing control parameters can be registered as increases or decreases in stroke volume. This approach is often applied during manipulation of the pAVI. A range of pAVIs is scanned, and the optimal interval is the pAVI that generates the largest increase in aortic VTI (stroke volume).

Figure 31-62 Doppler aortic velocity time integral for estimating stroke volume during manipulation of the pacemaker AVI.

Left, Signal acquisition is performed in the apical five-chamber view. Right, The aortic velocity time integral (AoVTI) envelope is outlined.

Figure 31-63 Effect of change in pacemaker AVI on aortic VTI.

Increase in pacemaker AVI from 80 to 150 msec increases aortic velocity time integral (AoVTI) from 16 to 23 cm. This is interpreted to indicate that pAVI 80 was “too short,” resulting in atrial truncation and reduced LV preload.

The aortic VTI method is simple to perform, assuming an adequate echocardiographic window can be obtained. Data acquisition is generally rapid, which allows for screening a broad range of pacing control parameters in abbreviated time. However, VTI measurements are influenced by extraneous factors (e.g., respiratory phase, sympathetic tone). The accuracy of the method is degraded because of at least 10% variation between measurements, and it cannot reliably discriminate changes in VTI smaller than this inherent error.

Noninvasive Hemodynamic Monitoring

Finger photoplethysmography (FPPG) has been used as a surrogate to invasive hemodynamic monitoring for optimizing AV resynchronization (Fig. 31-64). FPPG accurately tracks beat-to-beat changes in arterial pulse pressure at rest and during exercise and therefore provides an attractive alternative to direct LV pressure recordings. At rest, FPPG correctly identified positive aortic pulse pressure responses with 71% sensitivity and 90% specificity compared to invasive aortic pressure recordings, and the magnitude of FPPG changes in response to manipulation of the pAVI during CRT was well correlated with positive aortic pulse pressure changes (R² = 0.73; see Fig. 31-64).¹¹² However, the correlation with negative aortic pressure changes was poor (R² = 0.43). FPPG identified 78% of the patients having positive aortic pulse pressure changes in response to manipulation of the pAVI and identified the single pAVI that yielded the maximum increase in pulse pressure in all patients. Maximum increase in systolic blood pressure using FPPG has been used to identity the optimal pAVI during rest and exercise and duplicates the parabolic relationship observed in acute hemodynamic studies.¹¹³

Figure 31-64 Optimizing AV resynchronization using acute changes in aortic pressure recorded by finger photoplethysmography (FPPG).

A, Good correlation between positive changes in aortic pressure measured by FPPG compared to invasive measurement of aortic pressure (AOP). B, Optimal pacemaker AVI at rest and during atrial pacing just above sinus rate identified by maximum increase in systolic blood pressure using FPPG. C, Poor correlation between negative changes in AOP by FPPG as compared to invasive measurement of aortic pressure.

(Modified from Whinnet ZI, et al: The atrioventricular delay of cardiac resynchronization can be optimized hemodynamically during exercise and predicted from resting measurements. Heart Rhythm 5:378-386, 2008.)

Electrogram Analysis

The optimal pAVI for AV resynchronization can also be estimated using an EGM-based approach (Fig. 31-65). This is derived from timing observations made during invasive determination of the acute hemodynamic response (maximal positive change in LV +dP/dt) during manipulation of the pAVI. The single pAVI that generates the largest acute increase in LV +dP/dt is derived from the iAVI (time from local RA → local RV EGM) using two linear equations that incorporate baseline ventricular conduction delay (QRSd).¹¹⁴ If QRSd > 150 msec, the estimated optimal pAVI = 0.7*iAVI (msec) − 55 msec; if QRSd is 120 to 150 msec, the estimated optimal pAVI = 0.7*iAVI (msec). These regression formulas can be closely approximated by the following simple rules: the estimated pAVI for patients with QRSd > 150 msec is 50% of iAVI (P-R interval) and 75% for QRSd of 120 to 150 msec. These equations reflect the logical and necessary relationship between pAVI and ventricular conduction delay described earlier. This EGM method showed strong correlation (P <.0001) between maximum % increase in LV +dP/dt and maximum achievable % increase in LV +dP/dt during conditions of atrial sensing (R² = 0.99) and atrial pacing (R² = 0.96)¹¹⁵ (Fig. 31-66). This strategy for programming pAVI was employed in the Comparison of Medical Therapy, Pacing, and Defibrillation in Heart Failure (COMPANION) trial¹¹⁶ and has been automated in one device-based application (SmartDelay, Boston Scientific; see later).

Figure 31-65 Optimal AV resynchronization using electrogram analysis.

A, Surface ECG during intrinsic rhythm. P-R interval = 220 msec, QRSd (LBBB) = 170 msec. Note E-A fusion on mitral inflow caused by PR prolongation (e.g., iAVI is “too long”). B, Surface ECG during BiV pacing at estimated optimal pacemaker AVI = 100 msec, derived from timing of RA-RV EGM (200 msec) and QRSd 170 msec (see text for calculation). Note E-A separation and lack of A-wave truncation.

Figure 31-66 Atrial sensing (A) and pacing (B).

Correlation between % LV +dP/dt using the electrogram (EGM) method and maximum achievable % LV +dP/dt.

(Modified from Gold MR et al: A prospective comparison of AV delay programming methods for hemodynamic optimization during cardiac resynchronization therapy. J Cardiovasc Electrophysiol 18:490-496, 2007.)

Comparison of Methods for Optimizing AV Resynchronization

A direct comparison of the aortic VTI versus mitral inflow pattern method, using maximum increase in VTI as the endpoint, reported superiority of the aortic VTI method (58% greater relative increase, ~7% greater absolute increase in maximum LV stroke volume).¹¹⁷ The optimal pAVI predicted by the mitral inflow method was significantly shorter than the VTI method. A similar conclusion was reached in a comparison of the aortic VTI and mitral inflow methods using invasive hemodynamic measurements.¹¹⁸ In this study the maximum increase in LV +dP/dt showed good correlation with the mitral inflow (E-A) VTI (r = 0.96), poor correlation with the aortic VTI method (r = 0.54), and no correlation with the mitral inflow method (r = 0.35). It is argued that the aortic VTI method performs better than the mitral inflow method because it titrates a measure of LV ejection, whereas the mitral inflow method titrates a measure of LV filling. Using invasive hemodynamic monitoring, the EGM method was shown to yield the least difference between % LV +dP/dt and maximum achievable % LV +dP/dt during conditions of both atrial sensing and atrial pacing, compared with arbitrary pAVIs (100, 120, 140, 160 msec) or “best” pAVIs determined by the mitral inflow method or aortic VTI method.¹¹⁵

More importantly, no single method of AV resynchronization has demonstrated convincing clinical superiority, such as reduction in heart failure hospitalization or reductions in LV end-systolic volume (ESV) (reverse remodeling), although minimal prospective evaluation of any of these methods has been made. A randomized single-blind study of 40 patients compared the optimal pAVI determined by largest increase in aortic VTI against an empiric pAVI of 120 msec.⁸⁰ A small improvement in ejection fraction, NYHA functional class, and quality of life scores were observed in the VTI-optimized group. These differences were probably spurious, because the mean-VTI optimized pAVI and the empiric pAVI were almost identical (119 vs. 120 msec, respectively). Individual patient variation may have accounted for the slight differences in outcomes between groups.

Using variations on the mitral inflow method or invasive hemodynamic monitoring, the optimized pAVI is almost invariably in the range of 80 to 110 msec, regardless of other considerations.^{68,111,116,119,120} Therefore, some have argued empiric programming of the pAVI at about 100 msec. Further, the optimal pAVI varies with heart rate, loading conditions, and atrial pacing versus sinus rhythm.

Although ventricular activation fusion (electrical resynchronization) clearly is the translational mechanism of improvement in LV pump function (reverse remodeling), the additional contribution of optimal AV resynchronization to clinical outcomes after maximal correction of LV conduction delay has not been demonstrated.

Atrial Pacing/Pacemaker AVI Interaction: Loading Conditions and Diastolic/Systolic Function

Atrial pacing disturbs left-sided AV electromechanical coupling because of (1) latency in atrial capture and sensing, (2) interatrial conduction delay, (3) latency in ventricular capture, and (4) interventricular conduction delay. The situation is made even more complex because all four elements are influenced by atrial and ventricular lead positions. Capture latency refers to the delay between emission of the right atrial (RA) pacing stimulus and atrial contraction. Sensing latency refers to the delay between the onset of atrial depolarization and the time at which the local endocardial signal is sensed.

Because of latency in atrial capture and sensing, the optimal pAVI for sensed and paced P waves may differ. During sinus rhythm, the programmed pAVI begins when the native P wave is sensed but the physiologic A-V interval, which begins with atrial depolarization and the onset of mechanical contraction, may be delayed from sensing latency. The mean latency between the beginning of atrial depolarization and the time of atrial sensing is 30 to 50 msec. Thus, the physiologic A-V interval is longer than the programmed pAVI. The opposite situation occurs during atrial pacing. The programmed AV delay begins with emission of the atrial pacing stimulus, but the physiologic A-V interval begins with atrial depolarization and mechanical contraction. The mean latency between atrial output and capture is reported to be 30-50 ms, however it may be greater than 300 msec. The physiologic A-V interval is therefore shorter than the programmed pAVI. The magnitude of atrial capture and sensing latencies varies among patients and is influenced by many factors, including lead design, sensing circuitry, amplitude and rate of stimulation, and characteristics of the local endocardial atrial electrogram.

A further complication is the effect of autonomic innervation of the AV node during atrial pacing. Atrioventricular intervals during atrial pacing shorten during exercise and lengthen by as much as 150% in the standing or supine position.¹²¹ A mismatch between the lower pacing rate and autonomic balance may contribute to this situation. For example, patients with a relatively high lower rate programmed (70 or 80 bpm) often experience significant AVI extensions during atrial pacing while sleeping because of parasympathetic predominance.⁷

Additionally, significant interatrial conduction delays may result from cardiomyopathy and atrial enlargement, antiarrhythmic drugs, and other causes. Interatrial conduction delays are common during RA pacing and are influenced by pacing lead position. RA pacing from the regions of Bachmann’s bundle, fossa ovalis, and coronary sinus os results in less delayed left atrial (LA) activation compared to other common pacing sites, such as the RA appendage.

The common consequence of these effects is that the pAVI is already in progress before LA contribution to ventricular filling has begun. In this situation, the optimized pAVI during sinus rhythm may be “too short” during atrial pacing resulting in diastolic dysfunction.

This can be understood by analysis of atrial and ventricular activation times using the surface ECG and local electrograms (EGMs) (Figs. 31-67 to 31-73). P-wave duration represents total atrial activation time (AAT) (see Fig. 31-67). The earliest deflection of the P wave indicates the onset of RA activation (A). It is assumed that the latest termination of the P wave represents the end of LA electrical activation (B). Therefore, the AAT is measured as time interval A → B. Interventricular conduction delay (bundle branch block, BBB) results in sequential ventricular electrical activation. This is typically registered on the surface ECG as a “split” or “notched” QRS complex. The notch indicates the transition point between sequential monoventricular electrical activation. The portion of the QRS that occurs before the notch is composed of the first chamber activated, whereas the portion of the QRS that occurs after the notch consists of the second (delayed) chamber activated. The earliest first upstroke of the LBBB QRS indicates the onset of RV activation (right bundle branch, RBB) conduction (C in Fig. 31-67). The earliest first notch in the QRS indicates the transition point between the end of RV activation and the onset of LV activation caused by LBBB (D). The latest return of the QRS complex to baseline in any ECG lead indicates the termination of delayed LV activation from LBBB (E).⁷²

Figure 31-67 Analysis of atrial and ventricular electrical activation times on surface ECG.

See text for details.

Figure 31-68 Calculation of intrinsic atrial-ventricular electrical activation times on surface ECG.

See text for details.

Figure 31-69 Atrial-ventricular activation time by surface ECG and local EGM timing during sinus rhythm.

Figure 31-70 Differences in right-sided AV timing using local EGMs and left-sided AV electrical timing on surface ECG.

See text for details.

Figure 31-71 Effect of right atrial pacing on estimation of left-sided AV timing by local EGM timing.

See text for details.

Figure 31-72 Effect of right atrial pacing on left AV electromechanical timing during biventricular pacing (BVP).

See text for details. AVC, Atrioventricular conduction.

Figure 31-73 Effect of right atrial pacing on left AV electromechanical timing during BVP.

See text for details.

The time from P-wave onset to the first upstroke of the QRS (onset of RV activation from RBB conduction) indicates the right intrinsic AVI (iAVI) (A→C), which is the AV conduction time estimated by the P-R interval (see Fig. 31-68). The time from P-wave offset (LA activation offset) to the earliest notch in the QRS (LV activation onset) indicates the left iAVI (B→D). Time from QRS onset to first notch is RV activation time (RVAT, C→D); time from first notch to QRS offset is LV activation time (LVAT, D→E); time from QRS onset to offset is interventricular conduction time (C→E).

The effects of capture and sensing latency on local EGM timing relative to onset of electrical activity on the surface ECG are shown in Figure 31-69. The time from the earliest deflection of the P wave (RA activation onset) to the time of local RA EGM during sinus rhythm (atrial pacing-inhibited) is the RA sensing latency time (A→F). The time from earliest upstroke of the QRS (RV activation onset due to RBB conduction) to the time of the local RV EGM is the RV sensing latency time (C→G). The time from the local RA EGM to the local RV EGM is the right sinus rhythm pAVI, which is the EGM-based surrogate of the minimum right sinus rhythm iAVI (A→C). Therefore, at the atrioventricular level, because of RA and RV sensing latency, the right sinus rhythm pAVI and the true right sinus rhythm iAVI (indicated by surface activation timing) are not the same. The right sinus rhythm pAVI may be longer or shorter than the true right sinus rhythm iAVI. Similarly, at the ventricular level, the local EGM surrogate of the total interventricular activation time (C→E) is the time from local RV EGM to the local LV EGM (G→H). This inaccurately estimates true interventricular activation time because of (1) RV sensing latency, (2) LV sensing latency, and (3) variable timing of LV EGM relative to QRS offset, as determined by LV lead position. Also, because current CRT systems do not directly measure the offset of LV activation or the onset of earliest LV activation, right-sided EGM timing does not accurately reflect left-sided electromechanical activation (see Fig. 31-70). These effects are important because device timing instructions are EGM based, and EGM timing may be dissociated from the true onset and offset of electromechanical activation in the chamber of interest. This introduces the possibility of cross-chamber and multichamber timing errors that could negatively influence pump function.

The effects of right atrial pacing on multichamber timing can be understood with this framework (see Fig. 31-71). Atrial capture latency is the time from emission of the atrial pacing stimulus (AP) to the earliest deflection of the P wave (RA activation). Note the increase in AAT (A→B) during AP because of intra-atrial and interatrial conduction delay. The atrial paced pAVI is measured from the time the atrial pacing stimulus is emitted to the onset of native or paced ventricular activation. During native AV conduction, the RA paced AVI (pAVI) exceeds the true right AV conduction time by a factor equal to atrial capture latency and RV sensing latency. Note that the LA paced AVI (pAVI) is shorter than the left sensed iAVI (e.g., ↓B→D) because of the increased AAT during RA pacing, assuming no increase in AV conduction time (A→C). The consequence is that during RA pacing, the true right-sided AV electromechanical timing relationship is overestimated by local EGM timing, and the true left sided AV electromechanical timing relationship is unpredictably shortened.

The effects of RA pacing on left-sided AV electromechanical timing during biventricular pacing are shown in Figure 31-72. In this situation, the timing of ventricular contraction is controlled by the pAVI. The pAVI must be shorter than the time from RA activation to earliest LV activation (A→D), to advance delayed LV activation. Therefore, AP→G is the longest pAVI that will advance LV activation, whereas AP→B is the shortest permissible pAVI to prevent A-wave truncation. However, the values of AP→A, A→B, and G→D are not known and not incorporated into the pAVI. Consequently, because of atrial capture latency and increased AAT, the true left-sided AV electromechanical time is significantly shorter than the pAVI. The consequence of these effects is an increased risk that the pAVI is “too short,” resulting in left-sided timing disruptions (e.g., A-wave truncation). If RA pacing is accompanied by an increase in AV conduction time (see Fig. 31-73), the same considerations apply, except that the “outer bound” for the longest pAVI that will advance LV activation (AP→G) is increased. This may be a more favorable situation because a short pAVI that sufficiently advances delayed LV activation is less likely to result in left-sided timing disruptions (e.g., A-wave truncation).

The negative effect of atrial pacing on diastolic function during CRT is dramatic.¹²² The increase in AAT may result in severe diastolic dysfunction (A-wave truncation, E-A fusion) and reduced LV preload (Fig. 31-74). Typically, a pAVI offset (40-50 msec) is applied during atrial pacing to compensate for increased AAT by delaying ventricular contraction. In the absence of complete AV block, the increased pAVI results in progressive fusion with native ventricular conduction and withdrawal of ventricular resynchronization. The consequence of this unintended effect is a reduction in LV systolic pump function, indicated by increases in LV isovolumetric conduction time and ejection time. A sufficiently prolonged pAVI offset to prevent truncation of diastolic filling results in incomplete ventricular activation fusion. The probability of this scenario increases inversely with the P-R interval. At shorter P-R intervals during atrial pacing, pAVI extensions sufficient to eliminate A-wave truncation are more likely to result in loss of maximum LV preexcitation than at longer P-R intervals. A longer P-R interval provides greater leniency for allowing pAVI extensions to accommodate increased AAT without withdrawal of LV preexcitation.

Figure 31-74 Effect of atrial pacing on diastolic function during CRT.

Left, Timing periods of the cardiac electromechanical cycle (surface ECG, mitral inflow velocities, aortic outflow velocities). A, Atrial pacing during CRT in the DDD mode. Diastolic filling time is reduced because of A-wave truncation (indicated by ↓E-A duration). Isovolumetric contraction time (ICT) and ejection time (ET) are increased, indicating a reduction in systolic function. This is explained by unintentional partial withdrawal of ventricular resynchronization caused by the longer pacemaker AVI during atrial pacing intended to “offset” interatrial conduction delay. B, Improved diastolic and systolic performance in the VDD mode. Note ↑E-A duration and absence of A-wave truncation (resolution of diastolic dysfunction). Note ↓ICT and ↓ET caused by shorter pAVI, resulting in greater advancement of delayed LV activation and maximum ventricular resynchronization. Right, Diastolic function during different atrial pacing vs. sinus rhythm. A, DDD mode. Note E-A fusion caused by A-wave truncation despite longer pAVI “offset.” B, VDD mode. Note E-A separation in absence of right atrial pacing.

(Modified from Bernheim A, et al: Right atrial pacing impairs cardiac function during resynchronization therapy: acute effects of DDD pacing compared to VDD pacing. J Am Coll Cardiol 45:1482-1487, 2005.)

This problem is magnified by observations during invasive hemodynamic monitoring that indicate the maximum achievable increase in % LV dP/dt occurs at pAVI that is substantially longer during atrial pacing compared to atrial sensing (~mean 134 msec, range 69-262 msec vs. ~mean 208 msec, range 105-340 msec)^115,123 (Figs. 31-75 and 31-76). This mean difference is substantially longer than the typical pAVI offset during atrial pacing in most CRT pacing systems (~40 to 60 msec). As noted, unless such patients have significantly prolonged P-R intervals, pAVIs in this range will result in incomplete LV resynchronization because of fusion between native ventricular conduction and paced activation wavefronts. For this reason, AV junction ablation has been advocated from patients with (1) relatively short P-R intervals, (2) requirement for atrial pacing, and (3) requirement for long pAVI during CRT to overcome atrial pacing–related diastolic dysfunction. If AV conduction is completely eliminated, the pAVI during CRT necessary to prevent compromise of LV preload during obligatory atrial pacing is not restrained by progressive loss of ventricular resynchronization due to fusion with native ventricular conduction.

Figure 31-75 Optimal pAVIs during atrial sensing and atrial pacing using invasive hemodynamic monitoring (maximum increase % + LV dP/dt).

(Modified from Gold MR, et al: A prospective comparison of AV delay programming methods for hemodynamic optimization during cardiac resynchronization therapy. J Cardiovasc Electrophysiol 18:490-496, 2007.)

Figure 31-76 Example of greater maximal increase in % LV dP/dt during atrial pacing vs. atrial sensing but requiring longer pAVI. Note that the pacemaker AVI (283 msec) that yields maximum increase in % LV dP/dt during atrial pacing is about twice the pAVI (140 msec) that yields the maximum increase in % LV dP/dt during sinus rhythm.

(Modified from Gold MR, et al: Acute hemodynamic effects of atrial pacing with cardiac resynchronization therapy. J Cardiovasc Electrophysiol 20:894-900, 2009.)

Atrial pacing should be minimized or avoided altogether if possible. Programming approaches include reducing the lower pacing rate, eliminating sensor-modulated pacing (if active), or choosing an atrial-synchronous ventricular pacing mode that does not provide atrial pacing (i.e., VDD at a low programmed rate).

Implications of Spontaneous and Pacing-Induced Conduction Blocks During Biventricular Pacing

Ventricular conduction blocks present diverse and unpredictable challenges to generating ventricular activation wavefront fusion during CRT. Based on current limited understanding of the problem, the following conclusions can be made:

Against this background, a role of timed (sequential) ventricular stimulation exists. The ability to orchestrate ventricular chamber timing should theoretically provide additional leverage for overcoming functional and physiologic conduction barriers, to obtain maximum evidence of ventricular activation wavefront fusion. This is highlighted by the observation that simultaneous BiV pacing yields evidence of ventricular activation fusion, which is necessary for reverse volumetric remodeling, in only about 55% of cases.⁷²

Implementation of Sequential Ventricular Timing in CRT Systems

The ability selectively to manipulate the timing of RV and LV stimulation appeared in third-generation CRT systems. This arrangement requires separately programmable RV and LV stimulation outputs and circuitry to permit timing delay between outputs by stimulation site (V-V timing). The goal of V-V timing is site-selective, sequential ventricular stimulation. Theoretically, V-V timing could be achieved with unipolar or dual cathodal electrode configurations. From a practical perspective, unipolar pacing is inapplicable in CRT-D, and anodal capture at high outputs with dual cathodal electrode configurations results in unintended BiV stimulation and could disrupt V-V timing. This is probably only relevant when RV stimulation precedes LV stimulation, because RV capture will render the local myocardium refractory, and anodal capture during high-output LV stimulation will not occur. Accordingly, the use of V-V timing, where RV precedes LV stimulation with dual cathodal electrode configurations, mandates exclusion of anodal capture by the 12-lead ECG at programmed outputs (see earlier). Therefore, V-V timing is optimally delivered with true bipolar (RV and LV) electrode configurations.

Figure 31-95 shows an example of interventricular timing operation. The key operating points of V-V timing are (1) the pacemaker A-V interval determines the timing of the first ventricular pacing stimulus, (2) the pacemaker V-A interval (atrial escape interval) is calculated based on the timing of the first ventricular pacing stimulus, and (3) the timing of the second ventricular pacing stimulus is determined by the selected V-V interval. Some interactions with other aspects of pacemaker timing exist. For example, ventricular safety pacing and Ventricular Sense Response (Medtronic; see later) are delivered at the nominal V-V (~simultaneous BiV) interval. Interlocks between inviolable blanking periods, pAVI, ventricular tachyarrhythmia (VTA) detection intervals, and the selected V-V interval often arise and force reconciliation, usually in the form of more lenient spacing of sequential ventricular stimulation.

Figure 31-95 Interventricular (V-V) interval timing cycle operation.

Atrial sensing is shown on the left, atrial pacing on the right. The pacemaker A-V interval (pAVI) is initiated by either sensing (S) of the local atrial EGM (sinus rhythm) or the emission of an atrial pacing stimulus (P). An atrial blanking period (black bar) follows the sensed atrial event or atrial pacing stimulus. When the sensed pAVI (SAV) or atrial paced pAVI (PAV) expires, the ventricular pacing stimulus (P) for the first chamber selected (e.g., RV or LV) is emitted. In this manner, the programmed SAV or PAV controls the AV timing relationship for the first ventricular chamber stimulated. The first ventricular pacing stimulus initiates (1) a ventricular blanking period (black bar) and (2) the pacemaker V-A interval (atrial escape interval). The ventricular pacing stimulus (P) for the second chamber selected is emitted according to the programmer selected V-V interval (~4 to 100 msec). The second ventricular pacing stimulus must be delivered within the ventricular blanking period. Note the deformation of the evoked QRS complex by the second (delayed) ventricular pacing stimulus.

(Modified from Medtronic InSync CRT pacing system manual.)

Clinical Experience with Sequential Ventricular Stimulation

Despite the compelling rationale for timed ventricular stimulation during BiV pacing, the aggregate clinical experience has been a disappointment. Although acute studies demonstrated improvement in various metrics of LV systolic and diastolic performance during sequential versus simultaneous BiV pacing,^84,124,125 no such benefit has been recorded in randomized clinical trials. The Resynchronization for the Hemodynamic Treatment for Heart Failure Management (RHYTHM II ICD) study randomized 121 CRT patients to either simultaneous or “optimized” BiV pacing, possibly including sequential ventricular stimulation pacing.⁸² Echocardiographic metrics were used to determine V-V timing parameters in the “optimized” group. Interestingly, 28% were assigned simultaneous BiV pacing, whereas the remaining 72% received sequential ventricular pacing (V-V interval up to 80 msec). There was no difference in the composite endpoint of freedom from CRT-related complications, NYHA functional class, distance covered in 6-minute hall walk, or standard measures of quality of life at 6 months (Fig. 31-96). Similar absence of benefit of sequential BiV pacing was observed in the FREEDOM trial (see later).

Figure 31-96 Neutral effect of sequential vs. simultaneous BiV pacing.

There was no improvement in New York Heart Association (NYHA) functional class, distance covered in 6-minute hall walk, or standard measures of quality of life between sequential vs. simultaneous BiV pacing in the RHYTHM II ICD study.

(Modified from Boriani G, et al: Randomized comparison of simultaneous biventricular stimulation versus optimized interventricular delay in cardiac resynchronization therapy. Am Heart J 151:1050-1058, 2006.)

The failure to demonstrate a clinical advantage of sequential versus simultaneous BiV pacing could be explained by two opposing but related possibilities. First, simultaneous BiV pacing is sufficiently adequate in a substantial majority of patients such that a benefit of sequential ventricular pacing can only be demonstrated in a very large study sample. Second, failure to properly apply sequential ventricular stimulation negated the leveraging advantage for generating maximum ventricular activation wavefront fusion in a substantial majority of patients. The latter possibility seems most likely given considerations discussed previously. Without exception, comparisons between simultaneous and sequential ventricular pacing have been made using various echocardiographic metrics for titration of cross-chamber timing (AV and interventricular level). None of these approaches provides evidence that ventricular activation wavefront fusion, the translational mechanism and primary target of CRT titration, has been achieved. As described, the primary leveraging advantage of sequential ventricular pacing is the ability to overcome asymmetrical conduction delays across the left ventricle. An exemplary failure of echocardiographically directed application of sequential ventricular pacing is a recommendation for RV→LV stimulation in LBBB (e.g., wrong ventricular chamber first). This arrangement could theoretically yield paradoxical activation fusion without maximal correction of LV conduction delay if LV→RV conduction time is substantially less than RV→LV conduction time, a situation that may exist but has not been reported.

Methods for Automatic Adjustment of AV and VV Timing

Recognition of the importance of proper four-chamber electromechanical timing for maximum pump function justified attempts automatically to orchestrate timing adjustments at the atrial-ventricular (pAVI) and ventricular (V-V timing) level. Such control systems have been implemented in CRT pacing platforms with the common strategy of exploiting intracardiac EGMs to measure and adjust chamber timing.

QuickOpt (St. Jude Medical)

QuickOpt is an empirical EGM-based method that semi-automatically adjusts AV and VV timing. QuickOpt requires about 90 seconds to complete. Unlike SmartDelay, QuickOpt attempts to optimize LV filling (preload) by timing BiV stimulation to an EGM-based surrogate of IACT. This approach is substantially different from the SmartDelay, which uses the AVI to maximize LV global contractile function as measured by LV dP/dt.

QuickOpt Operating Specifics

QuickOpt has two adjustment stages: AV timing and VV timing. AV timing recommendations are made during forced atrial sensing on the basis of the time measured interval from RA sensed event (surrogate for P-wave onset) to the end of the far-field (RA tip to CAN) P wave, which is held to represent IACT (Fig. 31-97). The optimal iAVI and pAVI are then calculated according to the following formulas. When IACT > 100 msec, SAV_QO = IACT + 30 msec; when IACT < 100 msec, SAV_QO = IACT + 60 msec. These arbitrary offsets are intended to guarantee that ventricular pacing does not induce diastolic dysfunction (A-wave truncation, atrial transport block). The optimal pAVI_QO is similarly composed, except that an additional 50 msec is arbitrarily added to account for “atrial capture latency.” Limited justification for these calculations was provided in an unpublished clinical study of nine patients that demonstrated “reasonable correlation” with echo-based determination of optimal AVIs.¹³⁰

Figure 31-97 QuickOpt method for IACT estimate.

Interatrial conduction time (IACT) = onset of local RA EGM → offset of far-field P wave (RA electrode to CAN).

Sequential ventricular timing (VV) recommendations are made during forced BiV sensing and pacing. The intrinsic VV conduction time (RVs-LVs δ) is corrected for differences in paced VV conduction times. A correction factor ε is the difference in LV→RV conduction time (LVp-RVs) and RV→LV conduction time (RVp-LVs) determined during atrial-synchronous monochamber pacing at short pAVIs to guarantee 100% paced activation. The Optimal V-V interval is then calculated according to the following formula: VV_QO = 0.5 × (RVs-LVs δ + ε), where ε = 0.5 × [(LVp-RVs) − (RVp-LVs)]. If VV_QO = 0, simultaneous BiV pacing is delivered. If VV_QO > 0, sequential LV → RV pacing is delivered, whereas if VV_QO < 0, sequential RV → LV pacing is delivered.

Clinical Performance of QuickOpt

QuickOpt was evaluated in 58 patients using maximum aortic velocity time integral (AoVTI) as the surrogate hemodynamic endpoint¹³¹ (Fig. 31-98). High concordance was observed between maximum AoVTI determined by the echocardiogram method, compared with the AoVTI achieved at the EGM-based settings for optimal AVI (AV_QO), and optimal V-V interval (VV_QO). The conclusion of this acute surrogate hemodynamic study was that this EGM method for AV and VV timing adjustment was equivalent to an echocardiographic-based approach (AoVTI).

Figure 31-98 Concordance between maximum aortic VTI by echocardiogram method vs. EGM-based method using QuickOpt.

High concordance between maximal acute Doppler-derived and electrogram-derived aortic velocity time integral (AoVTI) during atrial sensed BiV pacing (top left), atrial paced BiV pacing (bottom left), and after VV optimization (top right).

The FREEDOM Trial (A Frequent Optimization Study using the QuickOpt Method) randomized 1647 patients to either periodic semi-automatic adjustment of AV and VV timing or standard of care-based optimization methods, including traditional echo-based optimization. Preliminary results, presented at the Heart Rhythm 2010 conference, demonstrated equivalent clinical outcomes between treatment arms, using a primary endpoint of heart failure clinical composite score (neutral, improved, worsened). In the RESPONSE-HF (Response of Cardiac Resynchronization Therapy Optimization with V-V Timing in Heart Failure Patients) trial, a higher percentage of CRT nonresponders became responders when QuickOpt optimization was used rather than simultaneous BiV pacing. The ongoing OPTIMISE CRT (Optimal Programming to Improve Mechanical Indices, Symptoms and Exercise in Cardiac Resynchronization Therapy) trial is testing the hypothesis that periodic semi-automatic adjustment of AV and VV timing will result in greater LV reverse remodeling than simultaneous BiV pacing.

Limitations of QuickOpt

Regarding the acute surrogate hemodynamic study of QuickOpt, the high concordance was calculated between the AoVTI at predicted optimal AVIs, not between the predicted optimal AVIs. This is expected given the relatively flat dose-response relationship between AoVTI on AV/VV timing adjustments. The accuracy of estimating IACT with RA near-field to CAN EGM has not been validated. This measurement can be easily confounded by lack of a clear offset of atrial activation on the far-field EGM (e.g., low-amplitude P wave) and far-field R-waves. The arbitrary offsets to the sAVI and pAVI are not validated, and this approach provides no guarantee that atrial truncation (transport block) does not occur. The physiologic basis for VV_QO equation is unknown. The purported goal of this approach is to generate ventricular activation wavefront fusion, but no evidence supports this. A key implicit assumption is that when intrinsic VV conduction delay and bidirectional paced ventricular conduction times are both 0, simultaneous BiV pacing will yield activation wavefront fusion. Another assumption is that wavefront fusion occurs with equal contribution (50%) from RV and LV pacing. Neither of these assumptions has been validated. Equivalent conduction times between widely spaced RV and LV stimulation sites does not guarantee that simultaneous BiV pacing will generate global ventricular activation wavefront fusion. None of these approaches provides evidence that ventricular activation wavefront fusion, the translational mechanism and primary target of CRT titration, has been achieved. QuickOpt recommended RV → LV sequential BiV pacing in 25% of patients with LBBB (e.g., wrong ventricular chamber first).¹³¹

Summary of Methods for Semi-Automatic Adjustment of AV and VV Timing

Semi-automated EGM-based adjustments to AV and VV timing are an expected evolution of CRT platforms. Such control systems are relatively easily implemented in devices systems at low overhead operating cost and have a natural clinical appeal. However, regardless of the specific technique, EGM-based control systems are immediately challenged by several critical shortcomings. The common flaw of all EGM-based approaches is the assumption that local electrical activation, timed by single-site EGMs, accurately depicts chamber level and cross-chamber level mechanical activation. There are a variety of explanations for this failure. Sensing and capture latencies can occur at all chamber levels. Consequently, mechanical activation may be underway before local activation exceeds the sensing threshold (sensing latency). Alternately, mechanical activation may be delayed after a pacing stimulus (capture latency). Both conditions result in timing errors in stimulation of a later-activated chamber. Right-sided pAVIs are flawed surrogates of left-sided electromechanical coupling because direct measurement of LA electrical activation is currently not feasible. Intra-atrial, interatrial, and interventricular conduction delays have unpredictable effects on left-sided AV electromechanical coupling.

Despite these limitations, it is probably true that local EGM timing provides a reasonable working approximation of cross-chamber electromechanical level at the (right-sided) AV level. In contrast, relative timing of single-point ventricular EGMs, or conduction times between two cross-chamber ventricular EGMs, cannot be assumed to accurately characterize global ventricular activation. Because the translational mechanism of CRT is activation wavefront reversal, the inability to characterize global activation sequence with current EGM-based timing measures is a serious deficiency. None of the current control systems is fully automated; optimal AV and VV timing requirements are dynamic, may change substantially over time, and probably require adjustment on virtually every cardiac cycle. Finally, none of the EGM-based approaches has been shown to improve clinical outcomes by any measure.