Electrical Activation During Sinus Rhythm

The cardiac action potential originates from the sinus node, located high in the right atrium (Fig. 9-1). Its cells depolarize spontaneously and initiate the spontaneous depolarization of action potentials at a regular rate from the sinus node. This rate depends on various conditions, such as atrial stretch and sympathetic activation, but is usually between 60 and 100 beats per minute (bpm) at rest. Myocytes are electrically coupled to each other through gap junctions. These structures consist of connexin molecules and allow direct intercellular communication.¹ Gap junctions do not have a preferential direction of conduction, but because the action potential starts in the sinus node, it spreads from there through the atria. Evidence suggests specialized conduction pathways in the atrium, but their (patho) physiologic relevance is still disputed.

Figure 9-1 Conduction of the impulse in the heart.

The impulse originates in the sinus node (1), continues in the atrial wall (2), and is delayed in the atrioventricular (AV) node (3). Conduction within the ventricles is initially rapid within the rapid conduction system: His bundle (4), right and left bundle branches (5), and Purkinje fibers (6). The impulse is transferred from the rapid conduction system to the working myocardium in the Purkinje-myocardial junctions (7), which are located in the endocardium. Within the slowly conducting working myocardium, the impulse is conducted from endocardium to epicardium.

In the human heart, spreading of the action potential through the atria takes approximately 100 msec, after which the impulse reaches the AV node (see Fig. 9-1). In the normal heart the AV node is the only electrical connection between the atria and ventricles, because a fibrous ring (anulus fibrosus) is present between the remaining parts of the atria and ventricles. The AV nodal tissue conducts the electrical impulses very slowly; indeed, it takes approximately 80 msec for these impulses to travel from the atrial side to the ventricular side of the AV node. This delay between atrial activation and ventricular activation has functional importance because it allows optimal ventricular filling. Its slow conduction renders the AV node sensitive to impaired conduction and even complete conduction block, an important indication for ventricular pacing. As with other locations in the heart, conduction in the AV node has no preferential direction. Consequently, impulses can also be conducted retrogradely through the AV node, a condition that can occur when the ventricles are electrically stimulated.

From the AV node, the electrical impulse reaches the His bundle, the first part of the specialized conduction system of the ventricles called the Purkinje system. Within this system the electrical impulse is conducted approximately four times faster (3-4 m/sec) than in the working myocardium (0.3-1 m/sec). This difference results because Purkinje cells are longer and have a higher content of gap junctions.^1–5

The intraventricular conduction system can be regarded as a trifascicular conduction system consisting of the right bundle branch (RBB) and two divisions of the left bundle branch (LBB). The RBB proceeds subendocardially along the right side of the interventricular septum until it terminates in the Purkinje plexuses of the right ventricle; the LBB also has a short subendocardial route.

It seems that the bifascicular vision of structure of the LBB is oversimplified.⁶ The general picture that now emerges is that the left ventricular Purkinje network is composed of three main, widely interconnected parts, depending on the anterior subdivision, the posterior subdivision, and a centroseptal subdivision of the left main bundle. This third medial or centroseptal division supplies the midseptal area of the left ventricle and arises from the LBB, from its anterior or posterior subdivision, or both. The fascicles continue in a network of subendocardially located Purkinje fibers. In the left ventricle, Purkinje fibers form a network in the lower third of the septum and free wall, which also covers the papillary muscles.^4,7 In humans the bundles are present only underneath the endocardium, whereas other species (e.g., ox, sheep, goat) have networks of Purkinje fibers across the entire ventricular wall.^8,9

It is important to note that the His bundle as well as the right and left bundle branches and their major tributaries are electrically isolated from the adjacent myocardium. The only sites where the Purkinje system and the normal working myocardium are electrically coupled are the Purkinje-myocardial junctions. The exits of the Purkinje system are located in the subendocardium of the anterolateral wall of the right ventricular (RV) and the inferolateral left ventricular (LV) wall^4,10 (see Fig. 9-1). This area of exit corresponds with the area of the ventricular muscle that is activated earliest.^2–5,11,12 The distribution of the Purkinje-myocardial junctions is spatially inhomogeneous, and the junctions themselves have varying degrees of electrical coupling.¹³

Endocardial activation of the right ventricle starts near the insertion of the anterior papillary muscle 10 msec after onset of LV activation¹¹ (Fig. 9-2). After activation of the more apical regions, the activation of the ventricular working myocardium occurs predominantly from apex to base, both in the septum and in the LV and RV free wall. Further depolarization occurs centrifugally from endocardium to epicardium as well as tangentially.^11,12 The earliest epicardial breakthrough occurs in the pretrabecular area in the right ventricle, from which there is an overall radial spread toward the apex and base. The last part of the right ventricle that becomes activated is the AV sulcus and pulmonary conus. Overall, the posterobasal area of the left ventricle (or an area more lateral) is the last part of the heart to be depolarized (see Fig. 9-2).

Figure 9-2 Three-dimensional isochronic representation of the electrical activation in an isolated human heart. Shaded scale indicates activation time in milliseconds.

(Modified from Durrer D, van Dam RT, Freud GE, et al: Total excitation of the isolated human heart. Circulation 41:899-912, 1970.)

The time between arrival of the impulse in the His bundle and the first ventricular muscle activation is approximately 20 msec,⁴ whereas total ventricular activation lasts 60 to 80 msec, corresponding to a QRS duration of 70 to 80 msec.¹¹ These numbers illustrate the important role of the Purkinje fiber system in the synchronization of myocardial activity. This role results from the system’s unique propagation properties and its geometrically widespread distribution. During normal orthodromic excitation, fast propagation over long fibers, together with wide distribution of Purkinje-myocardial junctions, induces a high degree of coordination between distant regions of the myocardium.

Mechanical Activation During Sinus Rhythm

As in many other muscle cells, electrical activation leads to contraction in cardiac myocytes, a process referred to as excitation-contraction (E-C) coupling. In cardiac E-C coupling, the calcium ion (Ca²⁺) plays a central role¹⁴ (Fig. 9-3). The contraction-relaxation cycle starts when depolarization leads to entry of Ca²⁺ into the cell through voltage-dependent L-type Ca²⁺ channels. This Ca²⁺ entry triggers a much larger amount of Ca²⁺ to be released from the sarcoplasmic reticulum (Ca²⁺-induced Ca²⁺ release). These processes increase intracellular Ca²⁺ from approximately 10⁻⁷ M to 2 × 10⁻⁶ M. This high calcium concentration catalyzes the interaction between myosin and actin filaments, leading to contraction. After repolarization, relaxation occurs as Ca²⁺ dissociates from the contractile apparatus and is taken up again by the sarcoplasmic reticulum through the action of sarcoplasmic reticular Ca²⁺–adenosine triphosphatase (SERCA). Intracellular Ca²⁺ homeostasis is also maintained through action of the sodium-calcium (Na⁺/Ca²⁺) exchanger (NCX). Normally, this exchanger removes Ca²⁺ from the cell (forward mode). Figure 9-3 also shows that the intracellular Ca²⁺ concentration rises rapidly after the upstroke of the action potential, but that there is some delay between the Ca²⁺ increase and the development of force. This delay is the main determinant of the delay between electrical and mechanical activation.

Figure 9-3 Excitation-contraction (E-C) coupling in myocardium.

A, Schematic representation of the major pathways involved in calcium ion (Ca²⁺) homeostasis in the cardiomyocyte. a, Ca²⁺entry via L-type Ca²⁺ channel; b, Ca²⁺-induced Ca²⁺ release from the sarcoplasmic reticulum; c, Ca²⁺-induced contraction; d, dissociation of Ca²⁺ from the contractile apparatus; e, reuptake of Ca²⁺ via sarcoplasmic reticular Ca²⁺–adenosine triphosphatase (SERCA); f, Ca²⁺ efflux via the Na⁺/Ca²⁺ exchanger (NCX). B, Schematic tracing of a Ca²⁺ transient, an action potential, and contractile force. The numbers indicate the different phases of the action potential: 0, upstroke; 1, fast early depolarization; 2, plateau phase; 3, repolarization phase; 4, resting membrane potential.

The electromechanical delay, that is, the delay between the depolarization and the onset of force development, amounts to approximately 30 msec.^15,16 On a global basis, this delay can be observed as the delay between the R wave of the electrocardiogram (ECG) and the rise in LV pressure (Fig. 9-4). Electromechanical coupling in failing hearts is different from that in normal hearts. A prominent feature of heart failure is that the relation between SERCA and the NCX changes. Often, SERCA is downregulated¹⁷ and NCX upregulated in the heart failure state.¹⁸ On the one hand, the reduced SERCA activity leads to less Ca²⁺ loading of the sarcoplasmic reticulum, resulting in less Ca²⁺ release during the subsequent activation, and thus a weaker contraction (systolic dysfunction). On the other hand, during diastole, Ca²⁺ removal from the cytosol is slower and incomplete, leading to a slower relaxation and greater diastolic stiffness (diastolic dysfunction). The upregulation of the NCX and the elevated intracellular Na⁺ concentrations in failing myocardium¹⁹ may compensate for the SERCA downregulation.^20,21 The high Na⁺ concentration facilitates the NCX to work in the “reverse mode,” that is, to remove intracellular Na⁺ and exchange it for extracellular Ca²⁺. This additional Ca²⁺ enhances systolic function at least to some extent. Because the SERCA pump is much faster in pumping calcium than the NCX, insufficiency in contraction and relaxation in failing hearts is most pronounced at higher heart rates. These changes result in decreasing contractile force with increasing heart rate, a negative force-frequency relation.¹⁷ The changes in the failing myocardium lead to the altered expression of other proteins involved in E-C coupling, as well as shifts in isoforms of various contractile proteins.²²

Figure 9-4 Events of the cardiac cycle.

Left atrial, aortic, and left ventricular pressures are correlated with electrical events and left ventricular volume. AC, Atrial contraction; AV, aortic valve; IC, isovolumic contraction; IR, isovolumic relaxation; MV, mitral valve; RVF, rapid ventricular filling. See text for description.

Because of the tight coupling between excitation and contraction, atrial activation is followed by atrial contraction, and ventricular activation by ventricular contraction. Consequently, atrial contraction precedes ventricular contraction, as illustrated in Figure 9-4. Because of this timing, the atrial contraction adds about 20% to the volume of the ventricles. This “atrial kick” increases the length of ventricular muscle cells and their sarcomeres. The sarcomere length is an important determinant of myocardial contractile force. Over the entire physiologic range of sarcomere lengths (1.6-2.4 µm), the longer the sarcomeres, the greater is the contractile force. This effect, known as the Frank-Starling relation (Starling curve; Starling law of the heart), is only partly related to the overlap of actin and myosin, as generally mentioned in textbooks. More important to the Frank-Starling relation is the growing affinity of troponin C for calcium at increasing sarcomere length,²³ although the cause of this greater sensitivity is not completely understood.

Ventricular contraction starts after ventricular depolarization. After a short isovolumic contraction phase, the aortic valve opens, starting the ejection phase. The velocity of emptying of the ventricle is highest in the first half of the ejection phase because of a higher pressure gradient between the left ventricle and the aorta. In the second half of the ejection phase, LV pressure actually falls below aortic pressure, but the aortic valve stays open because of the inertia of the flowing blood. With increasingly negative LV-aortic pressure gradients, the aortic valve closes and the isovolumic relaxation phase starts, ending with the opening of the mitral valve. Because of the filling of the atrium during ventricular systole, atrial pressure is relatively high. This event, in combination with the rapid fall in LV pressure, causes a positive AV pressure gradient during the early filling phase. Thus, rapid acceleration of blood occurs, contributing to most of the LV filling in the early diastolic phase. This event is also reflected by the large Ei wave on mitral valve Doppler recordings in healthy hearts (Fig. 9-5).

Figure 9-5 Pulsed Doppler recording of transmitral flow, illustrating the technique used to measure various segments of the timed velocity integral and diastolic filling period (ms, milliseconds; m/s, meters per second). The relationship between early diastolic filling and filling associated with atrial systole can be quantitated. Possible measurements include early and atrial velocities, the ratio of early to atrial velocities, the early (Ei) and atrial (Ai) flow-velocity integral (area under flow-velocity curve), the ratio of early to atrial flow-velocity integrals, and the amount of diastolic filling occurring in the first third of diastole ( DFT [diastolic filling time]).

(From Pearson AC, Janosik DL, Redd RM, et al: Doppler echocardiographic assessment of the effect of varying atrioventricular delay and pacemaker mode on left ventricular filling. Am Heart J 115:611, 1988.)

Atrial contraction, after the P wave on the ECG, produces a surge of blood into the ventricle at the end of diastole, causing the final and optimal filling of the ventricle. Reversal of the pressure gradient across the mitral valve, caused by atrial relaxation and ventricular contraction, pulls the valve cusps in apposition, facilitating closure of the valve. Therefore, coupling and proper timing of atrial and ventricular contractions are important determinants of ventricular pump function. The loss of atrial systole diminishes effective LV stroke volume by 25% at low heart rates and by almost 50% at higher heart rates. How different modes of pacing influence the atrial contribution to pump function and how this contribution affects pump function in different categories of patients are discussed later.

Although both Frank-Starling and force-frequency relations are properties of the myocardium itself, which can regulate cardiac function to an important extent, extrinsic factors also modulate cardiac performance. Many of these factors are related to the autonomic nervous and hormonal systems. The afferent parts of these systems can be divided into low-pressure and high-pressure domains. Low-pressure sensors are present in the atria and pulmonary circulation. Stimulation, from increased pressures and stretch, leads to parasympathetic stimulation and sympathetic withdrawal. Moreover, atrial stretch induces release of atrial natriuretic factor (ANF). This 28–amino acid polypeptide is a potent arterial and venous vasodilator that increases urine production. The low-pressure sensors monitor the filling status of the circulation. However, erroneous readings may occur during inadequate AV synchronization, leading to elevated atrial pressures (cannon A waves). Stimulation of the reflexes and ANF production are involved in the so-called pacemaker syndrome (see later).

The high-pressure sensors consist predominantly of the baroreceptors in the arterial system. These sensors feed back to the brainstem through afferent neurons in the vagal and glossopharyngeal nerves. Stimulation of these receptors through a decrease in blood pressure induces sympathetic stimulation and parasympathetic withdrawal. Although reduced parasympathetic activation is the most important factor leading to the increase in heart rate during hypotension, the greater sympathetic stimulation induces arterial and venous vasoconstriction and increases myocardial contractility.

Other pressure sensors are found in the juxtaglomerular apparatus in the kidneys. Their stimulation leads to greater production of renin, angiotensin, and aldosterone, resulting in vasoconstriction as well as water and salt retention. These systems are effective in maintaining equilibrium in blood pressure and cardiac output during short-term variations. However, sustained neurohumoral stimulation is an important factor in the long-term development of hypertrophy and the rise in filling pressure.

Abnormal Activation Sequence During Left Bundle Branch Block and Right Ventricular Pacing

The normal, physiologic, and almost synchronous sequence of electrical activation is lost in diseases affecting the ventricular conduction system, such as block of the left or right bundle branch²⁴ and the presence of an accessory pathway bypassing the AV node, as in the Wolff-Parkinson-White syndrome.²⁵ Additionally, ectopic impulse generation, occurring during ventricular pacing and extrasystoles,^26,27 leads to abnormal impulse conduction. Under all these circumstances, the impulse is conducted primarily through the slowly conducting working myocardium rather than rapidly through the specialized conduction system. As a consequence, under conditions of abnormal activation, the time required for activation of the entire ventricular muscle, expressed as QRS duration, is at least twice as long as that during normal sinus rhythm.

The extent of asynchrony (dyssynchrony) and the sequence of activation during abnormal conduction are determined by at least four myocardial properties, as follows:

Detailed studies on the three-dimensional spread of activation during ventricular pacing in canine hearts have been conducted since the 1960s.^12,26 The sequence of electrical activation in LBB block (LBBB) is similar to that during RV apex pacing. This sequence can be derived from the QRS configuration of the surface ECG^27,34,35 and from endocardial activation maps in experimental LBBB and RV pacing.³⁶ The similarity has lead to the use of AV sequential RV apex pacing as a model for “experimental LBBB.”^34,37,38

Until recently, information on impulse conduction in patients with LBBB was limited to the data reported by Vassallo et al.^24,27 These investigators mapped LV endocardial activation during RV apex pacing and in LBBB at a limited number of endocardial sites, showing that activation starts at the RV endocardium in both conditions. The first noticeable activation at the LV endocardium after right-to-left conduction of the impulse occurs at a single breakthrough site, which in almost all patients is the LV breakthrough time, 50 to 70 msec after the earliest RV activation. The impulse is conducted from the septum toward the distal free wall in a gradual manner, the site of latest activation generally being the inferoposterior wall.¹¹ More recently, this finding has been confirmed using high-resolution mapping system^39–42 (Fig. 9-6). These studies also showed that activation patterns differ among patients largely because of differences in the origin of LBBB: either a proximal conduction block^43,44 or a uniform but slow conduction through the LBB.⁴⁵ In approximately one third of patients with heart failure and typical LBBB QRS morphology, transseptal activation time (i.e., time between earliest RV and LV septal breakthrough point) is bimodally distributed and has a large range in both groups of patients, with transseptal conduction time of 40 msec or less and longer than 40 msec.⁴⁶ This indicates great heterogeneity in the RV-to-LV activation time. Furthermore, in patients with LBBB, the total LV endocardial activation time ranges from 60 to 160 msec. Etiology does not seem to have a major impact on the total endocardial activation time. Of note, the sum of transseptal and total endocardial activation times does not account for the maximum QRS duration, which is 20 to 60 msec longer, probably because of LV endocardial-to-epicardial conduction time (Fig. 9-7). These data also show that QRS duration provides a reasonable although imperfect estimate of total electrical asynchrony.

Figure 9-6 Contact and noncontact mapping in patient with left bundle branch block and heart failure. The unipolar isopotential activation sequence (upper panel) shows a U-shaped activation front that rotated around the apex and activated the lateral wall late, whereas the bipolar propagation map (lower panel) shows a longer activation time in the anterior region. ANT, Anterior; LAT, lateral.

(Modified from Auricchio A, Fantoni C, Regoli F, et al: Characterization of left ventricular activation in patients with heart failure and left bundle-branch block. Circulation 109:1133-1139, 2004.)

Figure 9-7 A, Distribution of transseptal time, total endocardial activation time, and total QRS duration in 140 heart failure patients with a QRS duration of 120 msec or longer. Patients are ordered according to and transseptal time. B, Three-dimensional electroanatomic mapping of the right ventricle (RV) and left ventricle (LV). A color-coded activation sequence indicated the earliest (red) and the latest (blue) endocardial activation region. The earliest activated region is the anterolateral region of the RV, and the latest part is the posterobasal region of the LV.

Patients with LBBB show a peculiar spread of ventricular activation.^40,42 Indeed, in two thirds of patients with LBBB, a functional line of block is present, resulting in a U-shaped electrical activation wavefront. The presence and location of the line of block have been also recently observed in canine model of LBBB using the same noncontact mapping system.^46a The activation front propagates first around the inferior wall and then spreads onto the lateral and basal walls. The location and length of the line of block vary greatly and are related to the site and time of LV breakthrough. In patients with significant prolongation of the QRS duration (>150 msec), the line of block is consistently located in an anterior position. In contrast, in patients with QRS duration of 120 to 150 msec, the line of block is usually shorter and is located anteriorly, laterally, or posteriorly⁴⁰ (Movie 9-1).

Abnormal Contraction Patterns During Left Bundle Branch Block and Right Ventricular Pacing

Given the tight relationship between excitation and contraction in the myocardium, it is not surprising that asynchronous electrical activation also leads to asynchronous contraction. Regions where the impulse arrives first also start to contract first.⁴⁷ During LBBB or ventricular pacing, local contraction patterns differ not only in the onset of contraction, but also, and more importantly, in the pattern of contraction. These contraction patterns imply that opposing regions of the ventricular wall are out of phase and that energy generated by one region is dissipated in opposite regions. In patients with LBBB or RV pacing, the early-contracting region is most often the septum. The earliest-contracting fibers can shorten rapidly by up to 10% just during the isovolumic contraction phase, because the remaining muscle fibers are still in a relaxed state (Movie 9-2). This rapid, early shortening is followed by an additional but modest systolic shortening, eventually followed by systolic stretch (from delayed mechanical contraction of other regions, in lateral wall), and premature relaxation (Fig. 9-8). In late-activated regions, in contrast, the fibers are stretched in the early systolic phase (up to 15%) from contraction of the early-activated region. Doubling of net systolic shortening and delayed relaxation occur in late-activated regions.^31,48 The discoordination between early- and late-activated regions leads to lower output and reduced efficiency of the heart as a pump.

Figure 9-8 Contraction patterns in 24 regions of the anterior septum and anterior left ventricular (LV) free wall during pacing at the LV lateral wall. Each graph in the array represents circumferential strain versus time in a specific region. The left ventricle is displayed as if cut down the septum and folding the surface out onto the page. For the sake of clarity, the posterior half of the left ventricle is not shown, but the distribution of strains in this part of the LV wall is a mirror image of that in the anterior wall. Rows represent different levels from base (top) to apex (bottom) of the left ventricle, and columns are circumferential positions from midseptum (far left) via the LV anterior wall to the LV lateral wall (far right). Tics at the vertical axis denote a strain of 0.1 (≈10% shortening); tics on the horizontal axis denote 100 msec. Note the early shortening near the pacing site (*) and the prestretch in regions remote from the pacing site.

(Modified from Prinzen FW, Peschar M: Relation between the pacing induced sequence of activation and left ventricular pump function in animals. Pacing Clin Electrophysiol 25:484-498, 2002.)

The regional differences in contraction pattern most likely results from the local differences in myocardial fiber length during the early systolic phase. This is supported by studies using two isolated papillary muscles in series in which asynchronous stimulation caused a downward shift in the force-velocity relation in the earlier-activated muscle and an upward shift in the later-activated one.⁴⁹ Furthermore, during ventricular pacing, regional systolic fiber shortening increases with greater isovolumic stretch.⁵⁰ Similarly, a close correlation exists between the time of local electrical activation and the extent of systolic fiber shortening.⁵¹ Therefore, the regional differences in contraction pattern during ventricular pacing are most likely caused by regional differences in effective preload and local differences in the contraction force triggered by the Frank-Starling relation.

On M-mode echocardiography, abnormal contraction patterns during RV pacing and LBBB appear as paradoxical septal motion.⁵² However, this motion is not actually paradoxical, because it is really the net result of different forces. The motion is caused by the asynchrony between RV and LV, which produces dynamic alterations in transseptal pressure differences,^52,53 and by presystolic shortening of septal muscle fibers.⁵⁴ Abnormal septal motion results in a diminished contribution of the interventricular septum to LV ejection. This has been measured in experimental LBBB using MRI tagging⁵⁴ and in patients with LBBB QRS morphology using speckle tracking.^55,56 The septal regions, even in absence of detectable scar, showed the lowest amount of systolic strain within LV; in contrast, the lateral wall shows the highest regional strain.⁵⁵ This finding demonstrates that LBBB determines a unique and unequal LV strain distribution that may be corrected by CRT.^55–57

Effect of Left Bundle Branch Block and Right Ventricular Pacing on Local Energetic Efficiency

The local differences in wall motion and deformation mentioned previously reflect regional differences in myocardial work.^48,58 This relationship was demonstrated by construction of local fiber stress–fiber strain diagrams and calculation of local external and total mechanical work. In regions close to the pacing site (or septum in patients with LBBB), shortening occurs at low pressure, whereas these areas are transiently being stretched at higher ventricular pressures. As a consequence, the stress-strain loops have a figure-8-like shape with a low net area, indicating the absence of external work. In regions remote from the pacing site (or lateral wall in patients with LBBB), the loops are wide, and external work can be up to twice that during synchronous ventricular activation (Fig. 9-9). Total myocardial work (sum of external work and potential energy) in LBBB and with RV pacing is reduced by 50% in early-activated regions and is increased by 50% in late-activated regions in comparison with atrial pacing.^48,54,59

Figure 9-9 Upper panels, Maps of local external work during normal (right atrial), left ventricular (LV) free wall, and right ventricular (RV) apex pacing in a dog heart. External work values are presented in color II scale; for values, see scale bar; asterisk denotes pacing site. The LV wall is represented as a circle with the base located at the outer contour and the apex in the middle. Lower panels, Fiber length–fiber stress diagrams from three regions are displayed, indicating the shape of the diagram in normally activated (left), early-activated (middle), and late-activated (right) regions.

(Modified from Prinzen FW, Hunter WC, Wyman BT, et al: Mapping of regional myocardial strain and work during ventricular pacing: experimental study using magnetic resonance imaging tagging. J Am Coll Cardiol 33:1735-1742, 1999.)

Several studies reported that ventricular pacing and LBBB are associated with regional differences in myocardial blood flow,^{31,35,38,54,58–60} glucose uptake,^38,61 and oxygen consumption.^58,62 During RV apex pacing and LBBB, low values are found in the septum, the early-activated part of the left ventricle. Compared with sinus rhythm, myocardial blood flow and oxygen consumption are 30% lower in early-activated regions and 30% higher in late-activated regions.^31,58,59 Controversy surrounds the cause of the reduced septal blood flow and glucose uptake in LBBB and RV apex pacing. Various observations, such as close correlations among local myocardial oxygen consumption, work, and blood flow, suggest that the lower blood flow is a physiologic adaptation to the lower demand. However, at higher heart rates, the asynchronous contraction pattern may also impede myocardial perfusion.^38,63 In patients with RV pacemakers, false-positive findings of perfusion defects may result from this redistribution.⁶⁴

The nonuniform contraction patterns translate into a lower efficiency of conversion of metabolic energy into mechanical pump function. This was observed in isolated, isovolumically beating hearts,⁶⁵ as well as in anesthetized open-chest,^66,67 closed-chest,⁵⁹ and conscious dogs.⁶⁸ Mills et al.⁵⁹ also showed that efficiency remained low after 4 months of RV pacing. In all these preparations, RV apex pacing reduced mechanical output, whereas myocardial oxygen consumption was unchanged or even increased compared with atrial pacing; thus efficiency decreased by 20% to 30% in these studies. The opposite change was found when ventricular activation in patients with preexisting dyssynchrony caused by LBBB-like activation was improved by biventricular (BiV) or LV pacing alone. In these patients, BiV pacing improved the maximal rate of rise of LV pressure (LV dP/dt_max) without increasing myocardial oxygen consumption, indicating improved efficiency.⁶⁹ Efficient conversion of metabolic energy to pumping energy is of particular interest in patients with compromised coronary circulation, because higher oxygen consumption over longer periods (as in β-agonist therapy) leads to higher mortality.

Effect of Asynchronous Activation on Systolic and Diastolic Pump Function

Both RV pacing and LBBB reduce systolic and diastolic function. These effects are independent of changes in preload and afterload. This conclusion is based on results from studies using preparations in which preload and afterload were controlled,⁶⁵ as well as those in which preload- and afterload-independent indices of ventricular function were determined^68,70–72 (Figs. 9-10 and 9-11). The negative mechanical effect of dyssynchronous activation has been observed under various loading conditions⁷³ and during exercise.⁷⁴ Ventricular pacing also causes a deterioration in pump function in patients with coronary artery disease,⁷⁵ as well as those with already-impaired LV function⁷⁶ (Fig. 9-12). Impaired regional and global cardiac pump function has been observed in patients and animals with LBBB, even if LBBB was not accompanied by other cardiovascular diseases.^53,54 Therefore, it appears that under all circumstances, dyssynchrony is an important, independent determinant of cardiac pump function.

Figure 9-10 Relationship between cardiac output and mean left atrial pressure (LA_mean) during atrial pacing (open circles) compared with ventricular pacing (solid circles) in an intact dog model. For any given mean LA pressure, the cardiac output is higher with atrial pacing than with ventricular pacing.

(From Mitchell JH, Gilmore JP, Sarnoff SJ: The transport function of the atrium: factors influencing the relation between mean left atrial pressure and left ventricular end diastolic pressure. Am J Cardiol 9:237, 1962.)

Figure 9-11 Pressure-volume (P-V) diagrams during sinus rhythm (SR) (purple lines) and left ventricular (LV) pacing (Pace) (blue lines) before and during caval vein occlusion. The end-systolic P-V relation (straight lines) shifts rightward during ventricular pacing. These data indicate that the left ventricle operates at a larger volume under all loading conditions, and that at a given LV volume, LV pressure development and stroke volume are reduced.

(Modified from Van Oosterhout MFM, Prinzen FW, Arts T, et al: Asynchronous electrical activation induces inhomogeneous hypertrophy of the left ventricular wall. Circulation 98:588-595, 1998.)

Figure 9-12 Hemodynamic effects of atrial and atrioventricular (AV) sequential pacing in patients with normal and impaired left ventricular systolic function. Note that +dP/dt (maximal rate of rise in LV pressure) is significantly lower in both groups of patients with AV sequential pacing than in patients with atrial pacing. However, only in patients with impaired LV systolic function −dP/dt (maximal rate of decline in LV pressure) is significantly lower, and the isovolumic relaxation time significantly longer, during AV sequential pacing than during atrial pacing.

(Modified from Bedotto JB, Grayburn PA, Black WH, et al: Alterations in left ventricular relaxation during atrioventricular pacing in humans. J Am Coll Cardiol 15:658, 1990.)

Although LV dP/dt_max is a preload-dependent parameter, it is also sensitive to changes in ventricular activation sequence, as shown both in animal models and in humans^34,76–78 (see Fig. 9-12). Thus, LV dP/dt_max appears to be an appropriate marker of dyssynchrony-induced changes in LV global contractile function and its correction by any pacing technique. Additionally, the rate of ventricular relaxation is slower during ventricular pacing, with more pronounced detrimental effect on relaxation, in patients with impaired ventricular function⁷⁶ (see Fig. 9-12). Isovolumic relaxation parameters, such as LV dP/dt_min and Tau, are strongly influenced by pacing.^78–80 Parameters of auxotonic relaxation (rate of segment lengthening or LV volume increase) are also lower during ventricular pacing than during sinus rhythm, but the difference is less pronounced than for the isovolumic relaxation parameters.^73,78

As a consequence of the slower contraction and relaxation, isovolumic contraction and relaxation phases last longer, thus leaving less time for ventricular filling and ejection.^53,65,81 Therefore, it is not surprising that cardiac output and systolic arterial and LV pressures are also affected by a dyssynchronous activation. In general, stroke volume is affected more than systolic LV pressure,^72,77 presumably because baroreflex regulatory mechanisms partly compensate the decrease in blood pressure. This idea is supported by the finding of higher catecholamine levels⁸² and greater systemic vascular resistance⁷⁴ during ventricular pacing. With regard to the changes in stroke volume, it is important to note that ventricular pacing, especially RV apex pacing, can induce mitral regurgitation, as has been demonstrated in animals^83,84 and patients.^85,86 In addition to reduced stroke volume at unchanged preload, ejection fraction is usually found to be depressed during ventricular pacing^74,87 as well as in LBBB.⁸⁸ Similarly, ventricular pacing can increase pulmonary wedge pressure.⁷⁴ The negative inotropic effect of ventricular pacing under various loading conditions is clearly illustrated by a rightward shift of the LV function curve, that is, the relationship between cardiac output and mean atrial pressure⁷¹ (see Fig. 9-10). Later studies showed a rightward shift of the end-systolic pressure-volume (P-V) relation (see Fig. 9-11), thus suggesting that for each end-systolic pressure, the LV must operate at a larger LV volume.^70,72,89

Electrical and Structural Remodeling

Asynchronous electrical activation not only has acute mechanical consequences, but also influences cardiac structure and function over time. A particular adaptation of the heart to ventricular pacing occurs when the abnormal activation created by pacing is stopped but repolarization remains abnormal, a phenomenon called cardiac memory.⁹⁸ Several investigators found evidence that, as in the nervous system, short-term cardiac memory (<1 hour) involves changes in ion channels and phosphorylation of target proteins,⁹⁹ and that long-term cardiac memory (≈3 weeks of pacing) involves altered gene programming and protein expression.^98,100 Interestingly, in humans, cardiac memory appears to reach a steady state within 1 week,¹⁰¹ which appears to be quicker than in dogs.⁹⁸ Costard-Jäckle and Franz¹⁰² have demonstrated opposite changes in repolarization in regions remote from and close to a pacing site starting within 1 hour of ventricular pacing.

The repolarization abnormalities related to cardiac memory also have a mechanical counterpart. During sinus rhythm immediately after a period of ventricular pacing, relaxation is disturbed.¹⁰³ Moreover, systolic function deteriorates between 2 hours and 1 week after ventricular pacing has been stopped (Fig. 9-13). After ventricular pacing is stopped for 1 week, it takes days for ejection fraction to return to its prepacing value.¹⁰⁴ It seems likely that changes in the function of an ion channel, such as the L-type calcium channel,⁹⁹ underlie the reduction in ejection fraction during the first week of ventricular pacing. Therefore, electrical and contractile remodeling appears to occur soon after the onset of asynchronous activation.

Figure 9-13 Left ventricular (LV) ejection fraction before pacing, during short-term and midterm (1 week) ventricular pacing, and during the first days after restoration of normal sinus rhythm. During the entire experimental period, the heart rate was kept at 80 bpm with the use of fixed-rate pacing at the right atrium and, during the week of ventricular pacing, dual-chamber pacing with an AV interval of 100 msec.

(Data derived from Nahlawi M, Waligora M, Spies SM, et al: Left ventricular function during and after right ventricular pacing. J Am Coll Cardiol 44:1883-1888, 2004.)

In addition to the long-term effect on cardiac memory, longer-lasting (>1 month) ventricular pacing and LBBB lead to major structural changes, such as ventricular dilatation and asymmetrical hypertrophy.^54,70,105 The ventricular dilatation appears related to the LV operating at a larger volume. Global hypertrophy may be induced by this global dilatation, as well as by the greater sympathetic stimulation⁸² (Fig. 9-14). The asymmetry of hypertrophy appears to be more prominent during LV pacing than during RV pacing and LBBB.^54,70,106 However, regardless of pacing site and underlying conduction disturbance, the asymmetry of hypertrophy is characterized by a more pronounced growth of the late-activated myocardium, that is, the same region that shows enhanced contractile performance because of early systolic prestretch. It has been observed in juvenile canine hearts that ventricular pacing leads to fiber disarray,¹⁰⁷ which may also be the result of the regionally different and abnormal distribution of mechanical work.^{31,48,58,59,108}

Figure 9-14 Possible relationships among the various consequences of asynchronous activation of the ventricles, resulting from ventricular pacing or conduction disturbances, and the deterioration of pump function over time. P-V, Pressure-volume. For details, see text.

In ventricles undergoing long-term pacing,⁷⁰ myocyte diameter is increased in late-activated regions, but the number of capillaries per myocyte is unchanged (van Oosterhout and Prinzen, unpublished data). Therefore, the diffusion distance for oxygen increases, potentially leading to compromised oxygenation, which is known to render hypertrophic myocardium susceptible to ischemia. Resynchronization has been shown to result in improved coronary reserve, suggesting that the structural remodeling at the microvascular level may also be reversible.¹⁰⁹ An alternative explanation is that CRT could be simply decreasing the oxygen demand in the late-activated regions.

In conditions such as infarction and hypertension, ventricular remodeling appears, at least initially, to be meant to compensate for the loss of function and the increased load, respectively. However, dilatation and hypertrophy do not reduce the asynchronous activation induced by pacing or LBBB; if any, they increase it.^54,70 The reasons are the longer path length of impulse conduction (dilatation) and the larger muscle mass to be activated (hypertrophy). Moreover, chronic asynchronous activation also reduces expression of gap-junction channels¹¹⁰ and induces lateralization of these channels.¹¹¹ Akar et al.¹¹² have related the reduced expression in nonfailing and failing asynchronous hearts to slower impulse conduction, especially in the late-activated regions. Moreover, they showed that action potential duration increases in the late-activated myocardium. These molecular changes may render the myocardium not only more asynchronous, but also more susceptible to arrhythmias.

Even more complicated molecular changes occur in animal models with combined asynchronous activation and heart failure, such as that caused by rapid ventricular pacing. Under these conditions, beta-adrenergic receptors are downregulated,¹¹³ as are many ion channels, such as most potassium and calcium channels, whereas other channels display a regional difference in expression, such as the L-type calcium channel.¹¹⁴ Moreover, in both patients with LBBB¹¹⁵ and animal models of dyssynchrony,¹¹³ several proteins related to or facilitating the apoptotic process show unique regional distribution, further confirming the hypothesis that abnormal electrical activitation sequence may contribute to remodeling in heart failure patients.

All these processes may lead to a vicious circle in which dilatation and hypertrophy further reduce LV pump function, either directly or by increasing asynchrony of activation (see Fig. 9-14). As is the case after myocardial infarction, initial compensatory hypertrophy may result in heart failure many years later. Clinical evidence for such an important role of asynchronous activation in the development of heart failure is discussed later.

Correction of Bradycardia by Pacing

Cardiac pacing is the only effective treatment for symptomatic bradycardia. Permanent pacemakers have been implanted since the early 1960s to prevent death or syncope caused by ventricular asystole. Although asystole is clearly to be prevented, bradycardia does not always immediately lead to pump failure. This fact can be derived from studies on young patients with congenital AV block. As a compensation for the bradycardia, these patients have enlarged hearts and an increased ejection fraction.^116,117 In a 10-year follow-up study, 9 of 149 patients demonstrated dilated cardiomyopathy.¹¹⁷

At the onset of acquired complete heart block, cardiac output is smaller than with normal heart rate.¹¹⁸ As a result, compensatory increases in sympathetic tone and end-diastolic ventricular volume occur, leading to higher atrial rates, enhanced ventricular contractility, and augmented stroke volume. Over the short term, the canine heart adapts to bradycardia by increasing diastolic volume and neurohumoral activation, which results in BiV hypertrophy^118–120 (Table 9-1). Although these adaptations initially manage to return cardiac output to the level before the onset of AV block, this compensatory effect usually fades after 2 months of heart block. After 4 months, contractility is back to baseline, but LV cavity volume and corrected QT interval (QT_c) continue to increase¹¹⁸ (Fig. 9-15); the latter increase is associated with an enhanced susceptibility to arrhythmias.¹¹⁹ If complete AV block persists for more than 4 months, congestive heart failure (CHF) ensues.¹²⁰ Development of CHF may originate from the ventricular dilatation and hypertrophy initiated early after the start of heart block.

TABLE 9-1 Hemodynamic Effects of Ventricular Pacing in Complete Heart Block (CHB)

Figure 9-15 Effect of atrioventricular (AV) block on left ventricular (LV) cavity volume (structural remodeling), LV contractility (contractile remodeling), and corrected QT interval (QT_c-time) (electrical remodeling) in canine hearts, relative to their values before onset of AV block. Note the gradual and continuing increase in LV cavity volume over time, the early-onset and maintained electrical remodeling, and the biphasic contractile remodeling.

(Modified from Peschar M, Vernooy K, Cornelussen RN, et al: Structural, electrical and mechanical remodeling of the canine heart in AV-block and LBBB. Eur Heart J 25 Suppl D:61-65, 2004.)

The importance of the duration of overload is confirmed by the observation that the frequency of CHF symptoms in patients with complete heart block correlates with the duration of heart block. CHF has been described during chronic complete heart block even in patients with normal ventricular function. Brockman and Stoney¹²¹ reported that in two thirds of their patients with complete heart block and CHF, symptoms were relieved by ventricular pacing alone, and no additional medical therapy was required. Breur et al.¹²² showed that, in patients with congenital heart block, implantation of a pacemaker resulted in less LV dilatation than in the untreated control group, in which the LV cavity gradually dilated. Similarly, in dogs with AV block, LV dilatation gradually progresses over time (see Fig. 9-15), whereas ventricular pacing readily reverses the LV dilatation and hypertrophy caused by AV block.⁹³ However, longer-lasting ventricular pacing in hearts with AV block may lead to LV dilatation and hypertrophy secondary to the effect of asynchronous activation. Two studies on young adults who received pacing for approximately 10 years showed a high preponderance of CHF and LV dilatation.^123,124 These data strongly suggest that in the patient with AV block, the short-term and long-term effects of RV pacing are different. Therefore, the decision whether or not to start pacemaker treatment using RV pacing or the choice of the pacing site should be made carefully (see later).

Early studies suggested that the maximal increase in cardiac output during ventricular pacing at rest occurs at rates between 70 and 90 bpm^125–127 (Fig. 9-16). Further rises in rates result in either no additional increase or a decrease in cardiac output accompanied by greater peripheral vascular resistance. In contrast, there is growing evidence that lower heart rates may be particularly beneficial. Indeed, in animal models it appears that bradycardia promotes angiogenesis.^128,129 Moreover, Lei et al.¹³⁰ demonstrated that reducing normal heart rates by 20% to 30% through the continuous infusion of alinidine in a post–myocardial infarction (post-MI) rat model increased basic fibroblast growth factor, its receptor, and expression of related proteins. Bradycardia in this model also increased capillary length density in the border zone of the infarct by 40% and in the remote zone by 14%. Bradycardia also increased arteriolar density in the septum by 62%. These changes translated to a 23% greater coronary reserve, a smaller increase in post-MI left LV volume, and a greater preservation of ejection fraction in the bradycardic animals.

Figure 9-16 Histogram of optimal ventricular pacing rates at rest as defined by cardiac outputs determined by dye dilution in patients with complete heart block.

(From Sowton E. Hemodynamic studies in patients with artificial pacemakers. Br Heart J 26:737, 1964.)

Although the relationship between optimal resting ventricular pacing rate and ventricular dysfunction has not been systematically evaluated, it is likely that different resting ventricular rates are required in patients with systolic and diastolic dysfunction. For example, Ishibashi et al.¹³¹ reported that in elderly hypertrophic patients, even 25% increases in heart rate, as induced by atrial pacing, significantly reduced cardiac mechanical efficiency. The investigators went as far as to suggest that coronary perfusion itself in elderly hypertensive hypertrophic patients is negatively affected by an increase in heart rate (Fig. 9-17). They recommended that in the treatment of elderly hypertensive patients, the control of the heart rate in addition to the control of blood pressure might be helpful in minimizing the occurrence of myocardial ischemia and might slow down the subsequent progression to heart failure. Since then, their recommended treatment has become common practice with the use of β-blocker therapy to prevent MI and heart failure.

Figure 9-17 The relationship between heart rate and external mechanical efficiency of the heart (measured as left ventricular work divided by myocardial oxygen consumption expressed as energy using a conversion factor of 2.06), for basal condition (closed circles) and different atrial pacing rates (open circles, 25% above basal heart rate; triangles, 50% above basal heart rate) in elderly hypertensive patients.

In addition to maintaining resting heart rate in the physiologic range, pacemaker therapy preferably allows the heart rate to rise during exercise. A rise in heart rate with greater workload improves exercise capacity. Over the years a number of sensors have been incorporated into VVIR pacemakers. Many investigators have compared the exercise hemodynamics between sensor-driven VVIR pacing and VVI pacing. The percentage improvements of heart rate and exercise duration during symptoms-limited exercise in the VVIR and VVI pacing modes from a number of larger series for currently available rate-adaptive systems showed variable but consistent increases of 69% in rate and 32% in exercise tolerance when patients received VVIR pacing. Notably, most of the early comparative series of patients given VVI and VVIR pacing are rather small and have mixed patient populations in terms of symptoms of heart failure and degree of ventricular dysfunction. The use of an atrial sensing electrode to synchronize ventricular stimulation with intrinsic sinus node discharge (atrial synchronous ventricular pacing–VAT) allowed the implementation of a more physiologic form of pacing. The addition of ventricular sensing capability to this pacing mode resulted in the development of the VDD mode as a combination of VAT plus VVI. Both VVI and VDD modes of pacing prevent syncope due to bradycardia, but conventional VVI units are unable to respond to exertion by raising the pacing rate to meet the demand for a greater cardiac output.

The importance of increasing heart rate to augment cardiac output with exercise has been clearly documented in several studies.^132–134 Because the improvement in exercise performance is caused predominantly by an increase in heart rate, it is expected that favorable hemodynamic results will be obtained by rate-adaptive systems. Consequently, the modern approach to the management of complete symptomatic heart block involves the use of pacemaker units that provide not only basic ventricular pacing support, but also adaptation of the pacing rate to physiologic needs.

A major concern, especially in compromised hearts, is that heart rate should not increase too much, because shortened diastolic filling time, reduced LV compliance at higher rates of ventricular pacing, and increased systemic vascular resistance may limit cardiac output.^116,126 Rowe et al.¹³⁵ found other reasons not to increase pacing rate too greatly. They examined cardiac output and coronary blood flow at low (47 bpm), intermediate (77 bpm), and high (117 bpm) ventricular pacing rates in patients with complete heart block. Cardiac output rose with an increase from low to intermediate pacing but diminished at the high rate. However, coronary blood flow and cardiac oxygen consumption rose progressively with increasing heart rates, thus indicating lower pumping efficiency at the high rate.

Optimal Atrioventricular Interval

Importance of Atrioventricular Synchronization

Basically, the role of the atrial contraction is to help maintain a laminar flow of blood from the venous system to the ventricles; this role is tightly integrated with the movement of the ventricles to create a smooth motion of blood across the system. To perform its role, the atrium changes its function from that of a conduit during early filling to that of a booster pump during atrial systole, to that of a reservoir during ventricular systole. These changes are closely related to longitudinal displacement of the anulus of the mitral and tricuspid valves along the long axis of the heart. Indeed, the anulus is displaced toward the apex of the ventricles during systole and toward the atria during diastole, thus significantly contributing to the smooth movement of blood from the atria to the ventricles. This displacement keeps the overall volume of the heart almost constant during systole and diastole.¹³⁶

An optimally timed atrial contraction (before isovolumic contraction phase) maximizes LV filling and thereby its output by virtue of the Frank-Starling relation. Too early atrial contraction (as in long PQ times and dual-chamber pacing with long AV intervals) causes a loss of the booster pump function of the atrium. Moreover, early atrial contraction may initiate early mitral valve closure, thereby limiting ventricular diastolic filling time. This observation can be explained by the importance of three factors collaborating to achieve optimal closure of the AV valves. First, the ending of transvalvular flow at the end of the atrial contraction makes the valvular leaflets approach one another. Second, at the beginning of ventricular contraction, the anulus of the AV valves contracts, as do the papillary muscles that hold the leaflets. Simultaneously, at the start of ventricular contraction, ventricular pressure rises above atrial pressure, and the valves close. When these factors are misaligned, an opportunity for diastolic and systolic mitral regurgitation is created.

In the early years of pacing, the electrical stimulus was selectively applied to the ventricle (ventricular single-chamber pacing, or V pacing). With V pacing, the contraction of atria and ventricles is uncoupled, leading to an atrial contribution to LV filling that varies from beat to beat. This also results in large beat-to-beat variations in stroke volume, systolic pressure, and other hemodynamic variables.¹³⁷ The introduction of sequential AV pacing resulted in more regular heart beats and improved hemodynamics in both animals^78,138,139 and patients.¹⁴⁰ With advances in pacemaker technology, appreciation of the importance of maintaining AV synchrony has improved, and programming of optimal AV intervals in patients with dual-chamber pacemakers is of great importance.¹⁴¹

Figure 9-18 summarizes the main consequences of an optimal AV timing. Because filling time is limited, especially at high heart rates, the atrial contribution to stroke volume is more prominent at high than at low heart rates.¹⁴²

Figure 9-18 Hemodynamic effects of programming optimal atrioventricular (AV) delay.

Patients with Normal Left Ventricular Function

The AV interval that maximizes resting cardiac output during dual-chamber pacing varies widely among patients, usually reported as 125 to 200 msec. In patients with pacemakers, cardiac output increased by 4% to 20% when the AV interval was lengthened from 0 to between 100 and 130 msec.^{140,143–145}

Optimal AV interval can be determined in most patients during dual-chamber pacing by means of various invasive and noninvasive measurements. Most often, optimal AV interval is assessed by LV outflow recording with Doppler echocardiography (Fig. 9-19). Doppler-derived cardiac outputs at varying AV intervals can be used to fine-tune the AV interval. In patients with preserved ventricular function, this method is highly reliable, is sensitive to small output changes, and has a low interobserver variability. The optimal AV interval determined with Doppler echocardiography correlates well with the optimal AV interval determined with radionuclide ventriculography. Factors that may influence the optimal AV interval in different patients are summarized in Table 9-2. In addition to interpatient variability, other factors may influence determination of optimal AV interval in the same patient, such as heart rate, paced or sensed atrial event, posture, and drugs.^146–150 Moreover, in a specific patient’s lifetime, there may be situations in which the optimal AV interval is different from the AV interval considered to be physiologic. For example, AV sequential pacing for complete heart block complicating an acute MI or after cardiac surgery yields optimal hemodynamics at AV intervals in the range of 80 to 120 msec. This may be related to a surge of intrinsic catecholamine levels or to administered inotropic drugs and reduced LV compliance in these acutely stressful situations.

Figure 9-19 Left ventricular outflow recording at atrioventricular delay interval (AVDI; also AV interval) of 50, 175, and 300 msec during AV sequential pacing at identical heart rates in the same patient. The largest flow velocity integral (FVI) reflecting ventricular stroke volume occurs at AVDI 175 msec. Significant decreases are noted in stroke volume at AVDIs 50 and 300 msec.

(From Janosik DL, Pearson AC, Buckingham TA, et al: The hemodynamic benefit of differential atrioventricular delay intervals for sensed and paced atrial events during physiologic pacing. J Am Coll Cardiol 14:499, 1989. Reprinted with permission of the American College of Cardiology.)

TABLE 9-2 Possible Factors Influencing Optimal Atrioventricular Intervals

Type	Variables
Interpatient	Atrial capture latency Atrial sensing latency Intra-atrial conduction delay Ventricular capture latency Intraventricular conduction delay Underlying myocardial disease Heart rate Catecholamine levels

Intrapatient

Heart rate

Paced or sensed atrial event

Catecholamine levels

Drugs

The programmed AV interval regulates the timing of right atrial (RA) and RV conduction, which in a normal heart without significant atrial electrical disease usually coincides with that of left atrial (LA) and LV conduction. From a hemodynamic point of view, LA and LV conduction is most important, followed by the left-sided AV interval. The physiologic left-sided AV interval depends on several parameters, including the programmed AV interval, latency in atrial capture and sensing, interatrial conduction time, latency in ventricular capture, and interventricular conduction time.¹⁵¹ Because of the complex relationship among these intervals (see Fig. 9-19), significantly different right-sided and left-sided AV intervals in the same patient and in different patients can be recorded. Among these times, the one that most influences the optimal left-sided AV interval is the interatrial conduction time, measured as the time from the atrial pacing spike to the atrial depolarization on the esophageal electrogram or to the A wave on Doppler echocardiography. Patients with near-normal interatrial conduction time (≤90 msec) derive the greatest hemodynamic benefit from programmable AV intervals of about 150 msec; in contrast, in those patients presenting with prolonged interatrial conduction time, the AV interval should be set at 200 msec or longer to maximize cardiac output. Of particular note, programming a short AV interval in patients with prolonged interatrial conduction delay may result in depolarization of the LA after LV mechanical activation. This may produce hemodynamic values equivalent to or worse than those seen in patients with VVI pacing alone. Pacemaker syndrome may also occur under these circumstances.

Latency in atrial capture and sensing may lead to different optimal AV intervals for sensed P waves and paced P waves (Fig. 9-20). The magnitude of atrial capture and sensing latencies varies among patients and may be affected by the lead and pacemaker circuitry characteristics, electrode position, tissue interface, amplitude and rate of stimulation, P-wave morphology, myocardial disease, electrolytes and other metabolic factors, and drugs. The optimal AV interval for a sensed P wave is about 30 to 40 msec shorter than the optimal AV interval for a paced P wave followed by a paced QRS at a similar heart rate (Fig. 9-21). However, differences between optimal AV intervals for a paced P wave and a sensed P wave may be as great as 100 msec in some individuals. New dual-chamber pacemakers incorporate programmable features designed to adjust the AV interval automatically in response to paced versus sensed P waves and in response to the atrial rate. It is important to recognize that the physiologic AV interval is never equivalent to the programmed AV interval for either a paced or a sensed atrial event. The physiologic PQ interval begins with the onset of the P wave; however, with a sensed P wave, the programmed AV interval begins when the native P wave is sensed (so with a delay compared with onset of atrial activation), and with a paced P wave, the programmed AV interval begins with the atrial pacing spike, which is followed by further atrial depolarization.

Figure 9-20 Programmed atrioventricular delay interval (AVDI) and effect of PQ interval.

Top, When a paced P wave occurs, the effective PQ interval is shorter than the programmed AV delay because of latency in atrial capture. Bottom, When a sensed P wave occurs, the effective PQ interval is longer than the programmed AVDI because of latency in atrial sensing.

(From Janosik DL, Pearson AC, Buckingham TA, et al: The hemodynamic benefit of differential atrioventricular delay intervals for sensed and paced atrial events during physiologic pacing. J Am Coll Cardiol 14:499, 1989. Reprinted with permission of the American College of Cardiology.)

Figure 9-21 Interatrial conduction and change in the timing of left atrial (LA) depolarization with pacing mode change from DDD to VDD. The time delay between right atrial pacing artifact and LA depolarization is 115 msec (top), resulting in an LA-to-ventricular sequence of only 35 msec. With mode change to VDD (bottom), the LA-to-ventricular sequence extends by 75 msec, resulting in an LA-to-ventricular sequence of 110 msec.

(From Wish M, Fletcher RD, Gottdiener JS, Cohen AI: Importance of left atrial timing in the programming of dual-chamber pacemakers. Am J Cardiol 60:566, 1987.)

The importance of variation in AV interval during exercise in patients given physiologic pacing who have normal ventricular function is still controversial. Because the major determinant of augmentation of cardiac output during exercise is heart rate, the contribution of atrial systole and relative timing of the atrial and ventricular contraction may diminish in significance. Studies analyzing the effects of variation in AV interval during exercise have yielded conflicting results.^152–156 There are two major limitations to these studies: the relatively small sample size of most and the use of a fixed AV interval during each exercise test, but a different AV interval between exercise efforts. There is evidence that a rate-adaptive AV interval, which is automatically decremented in response to an increased atrial rate, is beneficial to cardiopulmonary performance during exercise.¹⁵⁷ In patients with chronotropic incompetence and high-level AV block, Sulke et al.¹⁵⁸ performed randomized double-blind crossover assessments of rate-adaptive and different fixed AV interval settings during 2 weeks of normal activity and performance on an exercise treadmill in patients undergoing DDDR pacing. There was a subjective improvement in quality of life with rate-adaptive AV intervals compared with fixed AV intervals, and patients preferred the rate-adaptive settings (Fig. 9-22). Notably, the longest AV interval (250 msec) was least preferred and was associated with the highest symptom prevalence. Moreover, exercise duration was not significantly different in any setting in DDDR mode but was significantly reduced in DDD mode.¹⁵⁸

Figure 9-22 Exercise capacity () and patient’s perception of general well-being () during everyday activity at fixed AV delay settings of 125, 175, and 250 msec and rate-responsive AV delay (RR AVD) settings in DDDR mode. , P < .05; m, P < .03; *, P < .01.

(Modified from Sulke AN, Chambers JB, Sowton E. The effect of atrioventricular delay programming in patients with DDDR pacemakers. Eur Heart J 13:464, 1992.)

Patients with Depressed Left Ventricular Function

In general, relaxation is slower in failing hearts than in normal hearts. Consequently, less blood enters the ventricles during the rapid filling phase, as can be observed from lower E waves on Doppler echocardiograms. Therefore, failing hearts are more dependent on properly timed atrial contraction than normal hearts. This has been demonstrated in comparisons of patients with decompensated heart failure and compensated nonvalvular heart disease.¹²⁵ This knowledge clarifies why optimal timing of atrial and ventricular stimulation in patients with pacemakers is of special interest in patients with heart failure. Some investigators have shown that pacemakers can actually improve the coupling between atria and ventricles over that without pacing. This issue applies to patients with excessively long PQ times. In two studies of subjects with these findings, hemodynamic benefit was achieved with AV sequential pacing at a physiologic AV interval^159,160 (Fig. 9-23). Reducing the prolonged AV interval increased diastolic filling time and reduced diastolic mitral regurgitation (Fig. 9-24). These beneficial effects are striking, because these studies were performed before the era of resynchronization, and the RV apex was used as a ventricular pacing site, presumably the least preferable site from a hemodynamic point of view (see later).

Figure 9-23 Mitral flow-velocity curve and simultaneous left atrial (LA) and left ventricular (LV) pressure curves in a patient with severe LV dysfunction and long PR interval.

A, Atrial pacing with antegrade native conduction and long atrioventricular (AV) delay interval (also AV interval). There is an increase in LV pressure above LA pressure during atrial relaxation and middiastole (arrowhead), resulting in a shortening of diastolic filling time and the onset of diastolic mitral regurgitation (m/s, meter per second). The baseline cardiac output (CO) is 3 L/min (l/m). B, AV pacing at short AV delay interval of 60 msec. Diastolic filling occurs throughout all of diastole. Atrial contraction occurs simultaneously with LV contraction, resulting in a reduction in cardiac output and an increase in mean LA pressure compared with A. C, Atrioventrucular (AV) pacing at the optimal AV delay interval of 180 msec. There is now an optimal relationship between the mechanical LA and LV contractions, so that diastolic filling period is maximized and the mean LA pressure is maintained at a low level, resulting in a rise in LV diastolic pressure and improvement in CO to 5.2 L/min.

(From Nishimura RA, Hayes DL, Holmes DR: Mechanism of hemodynamic improvement by dual chamber pacing for severe left ventricular dysfunction: An acute Doppler and catheterization hemodynamic study. J Am Coll Cardiol 25:281, 1995.)

Figure 9-24 Surface electrocardiographic lead I, atrial electrogram (AEG), ventricular electrogram (VEG), and pulmonary capillary wedge pressure (PCW) recordings from a single patient during atrioventricular (AV) pacing (AV pace) and ventricular pacing (V pace) at 80 bpm with an AV interval of 150 msec. The cannon A wave is noted (arrow) on the PCW tracings.

(From Reynolds DW: Hemodynamics of cardiac pacing. In Ellenbogen KA [ed]. Clinical cardiac pacing, pp 120-161. Cambridge, Blackwell Scientific, 1992. Reprinted by permission of Blackwell Scientific Publications.)

The actual role of the atrial contraction in patients with dilated ventricles and low ejection fraction varies among individuals. The booster pump action of the LA is noticeable as a “shoulder” in the LV pressure tracing, and the LV does not start contracting immediately after atrial contraction (Fig. 9-25). If this plateau lasts too long, diastolic mitral regurgitation could occur. About 20% of the patients in heart failure studies for whom complete hemodynamic data are available show this type of diastolic LV pressure waveform.^161–163 Figure 9-25 shows that using an optimized AV interval, in this case with BiV pacing, results in coincidence of the peak of the booster action of the atrial contraction and LV pressure development. Auricchio et al.¹⁶⁴ demonstrated that the maximum improvement in aortic pulse pressure, independent of the site being paced, occurs when this coincidence is achieved.

Figure 9-25 The atrial electrogram, right ventricular electrogram, left ventricular (LV) pressure, and aortic pressure tracings of a patient with New York Heart Association (NYHA) Class III heart failure from the Pacing Therapies for Congestive Heart Failure (PATH-CHF-II) trial.¹⁴⁹ Note the effect of programming an AV interval of 128 msec in the biventricular device. With this AV interval, the ventricular pressure starts right at the peak of the pressure increase created by the booster action of the left atrium. This AV interval provides the maximum preload at the minimal mean LV diastolic pressure while maximizing the time available for diastole, thus enabling higher heart rate increases without compromising filling time.

Examination and calculation of cardiac output with Doppler echocardiography, as frequently performed in patients with dual-chamber pacemakers, is more problematic in patients with depressed ejection fraction in whom the value of the effect of pacing may be near the rate of uncertainty in the method.¹⁶⁵ Thus, the number of repetitions required to obtain a clinically meaningful and reproducible approximation of the optimal AV interval makes this method, as single assessment, almost impractical for patients with depressed ejection fraction. The most utilized echocardiographic parameters for optimizing AV interval in patients with heart failure include echocardiography-based inflow or preload evaluations (e.g., velocity-time integral of transmitral flow or EA duration)¹⁶⁶ and measures of systolic function using time-velocity integral at the LV outflow tract.^167,168 Comparison of these different methods with invasive LV dP/dt_max to optimize AV interval in patients with CRT showed that mitral inflow velocity-time integral proved most accurate. The formerly used Ritter method, based on calculating the optimal AV timing by analyzing filling patterns obtained during only two AV intervals, has been found to be accurate to optimize AV interval in CRT patients.¹⁶⁹ Indeed, this method was created for patients with high-degree block, and some caution is advised in patients with intact AV conduction.¹⁷⁰ In these patients the electromechanical delay is not constant and varies as a function of the extent of collision between the intrinsically conducted depolarization wavefront and the wavefront generated by the artificial pacing spike. Therefore, in patients with an intact conduction system, the extent of collision between these two wavefronts will be different at the long and short AV delay intervals tested in the Ritter method, making the assumption required by the method completely invalid—that is, the electromechanical delay is unchanged at any tested AV delay.

Another consideration is important when deciding whether to use an inflow-based method for optimization of the AV interval. In patients with ventricular conduction delay who have CRT devices, the echocardiographic inflow-based method of optimizing AV delay interval neither optimizes the synchrony of the ventricular contraction nor accounts for how much preexcitation of the late-activated region is required to restore intraventricular synchrony (see later).

Clinical Consequences of Asynchronous Activation

The combination of acute adverse hemodynamic effects and long-term ventricular remodeling (see Fig. 9-14) may explain why abnormal electrical activation and asynchronous electrical activation have major implications for a patient’s clinical status. Studies have shown that morbidity and mortality are higher in patients with long-term RV apex pacing than with atrial pacing.^{145,171–173} In patients with sinus node disease and good ventricular function, the risk of heart failure was significantly high after more than 7 years in a comparison of atrial pacing and RV pacing.¹⁷³ In a similar population the risk of hospitalization for heart failure within 3 years increased with percentage of time patients underwent pacing at the RV apex.¹⁷⁴ Interestingly, development of heart failure and atrial fibrillation was more sensitive to percentage of pacing than to pacing mode (single- or dual-chamber pacing). In patients who received an ICD, the incidence of heart failure was higher within a year in those with pacing at VVI 70 bpm rather than in the backup mode.¹⁷⁵

Although pacing and, even more, heart failure started at a later age in the previous studies, RV pacing in children can also induce impaired pump function. Tantengco et al.¹²³ observed such adverse effects after approximately 10 years of pacing. The dramatic effects of ventricular pacing in otherwise healthy myocardium may be explained by the pronounced structural changes in juvenile hearts.^107,176 The deleterious effect of ventricular pacing is relatively easy to discover when start of pacing is exactly known, although spontaneous evolution in a dilated cardiomyopathy cannot be excluded. Whether adverse effects of asynchronous activation are also applicable to LBBB has been less clear, because LBBB is a silent event that is often accompanied by considerable comorbidity.^177–179

Longitudinal studies determining cardiovascular mortality and morbidity show that the presence of LBBB always carries a poor prognosis. In a 29-year follow-up study of 3983 pilots, the morbidity and cardiovascular mortality rate among those showing signs of LBBB was 17.2%, and the most common clinical event observed was sudden death without any previous symptoms (17%). These percentages are 10 times higher than those in subjects without LBBB.¹⁸⁰ In the Framingham Study, cumulative cardiovascular mortality over 10 years was approximately five times higher in patients with LBBB than in those without LBBB.¹⁸¹ In a population of 110,000, the risk for development of cardiovascular diseases was 21% in the 112 patients with LBBB, compared with 11% in age- and gender-matched controls. Moreover, cardiac mortality was strongly increased in the patients with LBBB.

The first direct evidence for an effect of LBBB on pump function came from studies in patients with intermittent LBBB.^182,183 Further proof of the negative effect of LBBB on hemodynamics has come from later studies in an animal model of LBBB, created by ablation of the proximal part of the LBB.^54,184,185 After ablation, duration of the QRS complex, asynchrony of RV-LV contraction, and asynchrony of electrical activation within the LV all increase significantly; also, relative to normal conduction values in the same species, these increases are similar in degree to the increases seen in patients with LBBB.^24,186 This finding suggests that, although the LBBB in patients may not be caused exclusively by proximal lesions, this model closely resembles the LBBB in patients. All studies on this animal model of LBBB show that LBBB reduces systolic and diastolic LV function and induces paradoxical septal wall motion.^54,184,185

Therefore, both RV apex pacing and LBBB seem to be conditions that increase the risk of heart failure and cardiac death, especially in patients with already-compromised function. For this reason, efforts to prevent RV apex pacing or correct LBBB are increasing. With current knowledge of electrical impulse conduction, the best solution may depend on whether the heart has normal or disturbed intrinsic conduction within the ventricles. In the following discussion, various possible strategies are described, with current and possible future applications. Strategies to maintain synchrony of activation are distinguished from those to restore synchrony.

Strategies to Maintain Ventricular Synchrony

Approaches to achieve or maintain ventricular synchrony include atrial pacing and pacing from alternative ventricular sites, including the His bundle. More recently, LV pacing and BiV pacing have been specifically indicated in patients with ventricular conduction disturbance. Both methods are commonly referred as “cardiac resynchronization therapy” and are discussed separately later.

Alternative-Site Ventricular Pacing

Patients with disturbed AV conduction are pacemaker dependent and require ventricular pacing. His bundle pacing is the way to maintain physiologic ventricular impulse conduction. Animal studies showed that QRS duration was shorter with pacing in the high ventricular septum than with RV apex pacing.^77,176,192 In the 1960s, Scherlag’s group showed that His bundle pacing results in the same QRS duration and pressure development as sinus rhythm and atrial pacing and in better hemodynamics than RV apex pacing.^139,193 In 2000, Deshmukh et al.¹⁹⁴ first reported successful His bundle pacing in patients on a permanent basis. Using improved implant tools, Zanon et al.¹⁹⁵ showed that His bundle pacing preserves normal distribution of myocardial blood flow and mechanics.

Nevertheless, because of the relatively difficult positioning of the pacing lead into the His bundle, pacing in the vicinity of His bundle is more often used. The literature refers to this position as “high RV septal” or “RV outflow tract” (RVOT), but in some studies, “RVOT” appears to imply pacing at the high RV free wall. Pacing in this region leads to abnormal QRS configuration compared with atrial pacing. Comparative studies of the acute hemodynamic effects of high septal pacing and RV apex pacing showed a moderate beneficial effect of septal pacing on LV pump function, although several studies found no significant difference between high septal and RV apex pacing.¹⁹⁶ Combined RV apex and RVOT pacing in patients resulted in shorter QRS duration than RVOT pacing alone but did not lead to further improvement of cardiac output.¹⁹⁷

Longer-term studies appear to show a more consistent beneficial effect of high septal pacing. Tse et al.¹⁹⁸ randomly assigned 24 patients to either RV apex or high septal pacing. After 18 months of pacing, perfusion defects and regional wall motion abnormalities were less common and ejection fractions were higher in the high septal pacing group. The reports explicitly mentioned that the “RVOT” lead was positioned in the interventricular septum. Unfortunately, however, the exact best site at the RV septum is not well known. One animal study showed that the RV site for best hemodynamic function differs for individual hearts.⁹⁴ Also, RV midseptal pacing performed similar to RV apical pacing.⁵⁹

In contrast, single-site LV pacing often results in better cardiac pump function than RV pacing.^91,199 Turner et al.²⁰⁰ showed that LV pacing may also lead to more synchronous activation if pacing is performed at longer AV intervals during maintained AV conduction in patients with a relatively narrow QRS complex. Because electrical asynchrony and mechanical asynchrony within the LV wall are similar in RV pacing and LV pacing²⁰¹ (Fig. 9-27), the less detrimental hemodynamic effect of LV pacing may be explained by better interventricular coupling. This mechanism may also partially account for the beneficial hemodynamic effects of LV pacing in patients with heart failure and AV node ablation.²⁰² In these patients, no fusion can occur between the impulse wavefronts from endogenous conduction and LV stimulation. However, long-term comparative data in some subgroups of patients are still missing, so no definitive conclusions can be drawn.

Figure 9-27 Three-dimensional reconstruction of left ventricular wall with myocardial strain represented in color.

Data were obtained using magnetic resonance imaging (MRI) tagging in a normal dog heart. Blue indicates contraction (negative circumferential strain), red indicates the reference state (end diastole), and yellow indicates stretch (positive circumferential strain). Data are shown for late diastole and early systole, midsystole, and late systole for right atrial (RA), biventricular (RVa+LV), RV apex (RVa), and left ventricular (LV) pacing. White arrow points to the midseptum. Note the considerable strain differences during both RVa and LV pacing in early systole and midsystole. RVa + LV pacing reduces these differences, but differences are still more pronounced than during RA pacing.

(Modified from Wyman BT, Hunter WC, Prinzen FW, et al: Effects of single- and biventricular pacing on temporal and spatial dynamics of ventricular contraction. Am J Physiol 282:H372-H379, 2002.)

Animal studies have shown that within the left ventricle, the septum and apex may be ideal pacing sites; pacing in these sites was associated with LV pump function close to that during normal sinus rhythm, even with pacing at short AV intervals^94,203 (Table 9-3). Studies in children now support the superior performance of LV apex pacing over RV pacing with respect to hemodynamic function⁹³ and echocardiographic indices of dyssynchrony²⁰⁴ and reverse remodeling.²⁰⁵ LV septal pacing is not yet clinically feasible, because positioning a lead in the LV cavity carries a risk of embolization and stroke.²⁰⁶ In an animal study, Mills et al.⁵⁹ advanced a lead from the right ventricle through the interventricular septum to reach the LV septal endocardium. Even after 4 months of pacing, LV septal pacing provided cardiac pump function and efficiency similar to that during natural conduction.

TABLE 9-3 Hemodynamic Parameters during VDD Pacing at Various Ventricular Sites in Dogs*

The degree of asynchrony induced by single-site ventricular pacing can be reduced by pacing at two or more sites simultaneously, preferably opposite to each other. This “multisite” pacing, however, reduces QRS duration by no more than 20% compared with single-site pacing.⁵⁰ Animal studies showed that BiV pacing does lead to better LV pump function than RV pacing.^50,94,207 Also, in patients with atrial fibrillation and AV node ablation, hemodynamic performance was better during BiV pacing than during single-site LV or RV pacing.²⁰⁸

It is important to note that in patients with normal ventricular conduction, such novel pacing sites (LV apex, RVOT, His bundle, multisite) at best prevented a reduction of pump function compared with atrial pacing or sinus rhythm, but never improved pump function. Also, left intraventricular mechanical asynchrony is only partially normalized, even with BiV pacing.²⁰⁹ Multisite pacing cannot outmatch the quick impulse conduction and its extensive spread through the network of Purkinje fibers. Therefore, avoiding any ventricular pacing would be preferable in hearts with normal ventricular conduction.

Minimizing Ventricular Pacing

One approach to avoiding any type of ventricular pacing has been to program fixed, long AV intervals; however, this approach resulted in occasional retrograde AV conduction and higher risk of arrhythmias in one third of a select population.²¹⁰ For many years, some pacemakers have incorporated an algorithm based on the programming of two AV delays: a short, physiologic AV delay and a longer AV delay. These devices start pacing with the longer AV delay and continue at this setting if they find intrinsically conducted ventricular beats. If the intrinsic conduction is not present or fails to occur, the devices automatically switch to the shorter and more physiologic AV delay. When functioning at the shorter AV delay, these devices periodically extend the AV delay to test for intrinsic conduction at programmable intervals. This type of algorithm is called AV search hysteresis.²¹¹

Rather than prolonging the AV delay, an algorithm called managed ventricular pacing (MVP) has been implemented in two studies.^212,213 With this algorithm, when intrinsic activity is present, only the atrium is paced; the ventricle is monitored on a beat-to-beat basis to verify intact AV conduction. Only in cases of transient and persistent loss of AV conduction does the MVP algorithm pacemaker switch to the DDDR mode. The mode intermittently tests for return of normal AV conduction. During atrial fibrillation, the device operates in the DDIR mode. With this approach, the cumulative percentage of ventricular pacing was reduced from a mean of 74% to 4%, with 80% of the patients being paced on the ventricle for less than 1% of the time. In a study with 1065 patients, MVP reduced the risk of developing persistent atrial fibrillation from 12.7% to 7.9% over a mean follow-up of 1.7 years.²¹⁴

Because many patients undergo pacing for sinus node disease, algorithms such as AV search hysteresis and MVP are promising for maintenance of ventricular synchrony in patients with predominantly intact AV and intraventricular conduction (Fig. 9-28). However, a recent study indicates that conservative programming of lower rate and rate response is advisable to minimize risk of AV decoupling and uncoupling as well as ventriculoatrial (VA) coupling.²¹⁵

Figure 9-28 Flow diagram showing most physiologic mode and/or site of pacing.

Potential adverse effects of ventricular pacing and conduction disturbance, such as left bundle branch block (LBBB), on cardiac pump function. Ventricular pacing can induce nonphysiologic atrioventricular (AV) as well as interventricular (biventricular, BiV) and intraventricular asynchrony. Assuming normal PQ times, LBBB causes only abnormal interventricular and intraventricular asynchrony. Effects of AV asynchrony are explained in text; LV, left ventricular; MVP, managed ventricular pacing; RVA, right ventricular apex.

Physiology of Cardiac Resynchronization Therapy

With the introduction of CRT, the use of cardiac pacemakers has been transformed from “just” artificial maintenance of the heart rhythm to restoration of pump function. Although the primary objective of BiV pacing may be to reduce intra-LV asynchrony, it can also be used to establish optimal interventricular coupling as well as AV coupling. The latter especially applies to left-sided AV coupling, because in LBBB the atrio-LV delay is usually prolonged. Parsai et al.²²⁴ showed that patients respond to CRT if at least one of the mechanisms of CRT is active.²²⁴

The predominant intraventricular conduction abnormality in about one third of patients with heart failure is LBBB, where activation spreads from the RBB to the RV wall and, after transseptal conduction, within the LV from the septum to the LV lateral wall (Fig. 9-29). Thus, pacing at the LV lateral wall is used to create an activation wavefront, which moves in the opposite direction from a spontaneously occurring activation wavefront.^16,26,201 Consequently, during BiV pacing, two activation wavefronts are generated and merge approximately in the middle.^36,209 A similar effect is obtained by single-site LV pacing using an AV interval that allows merging of the intrinsic activation originating from the RBB with the wavefront derived from the LV pacing lead. This merging wavefront leads to less electrical asynchrony in comparison with LBBB.³⁶ Because of the tight E-C coupling in the heart, CRT also improves coordination of contractions between the cardiac chambers and within the LV wall, as shown with different imaging techniques^55,225 (see Fig. 9-27). More specifically, CRT uniformly prolongs the time to maximum contraction⁵⁵ and thus restores a more coordinated contraction pattern, resulting in a more homogeneous distribution of regional loading conditions.⁵⁶

Figure 9-29 Color-coded electroanatomic isochronal maps of right ventricular (RV) and left ventricular (LV) activation in three patients with heart failure and left bundle branch block. The earliest ventricular activation site is recorded at the RV anterolateral region (red spot) in all three cases. After almost 40 to 60 msec, a single LV septal breakthrough site is noted (not shown). The latest-activated region is the LV posterolateral wall (blue to purple isochronal lines). The total activation endocardial time and endocardial RV and LV activation times are shown for each patient.

The improved synchrony leads to improvement of cardiac pump function, as determined by LV dP/dt_max, pulse pressure, cardiac output, and ejection fraction.* Such improved systolic pump function is achieved at unchanged or even lower filling pressures, denoting a true improvement of ventricular contractility through better coordination of contraction. Moreover, Nelson et al.⁶⁹ showed that better coordination of contraction improves mechanical pump function while slightly decreasing myocardial energy consumption, suggesting that CRT improves the efficiency of the cardiac pump. Further improvement in pump function may be mediated by a reduction in mitral regurgitation²²⁵ and prolongation of diastolic filling time. These beneficial effects occur almost immediately after the start of resynchronization.^36,161

Another possible mechanism for the improvement in LV pump function by BiV and LV pacing is that it improves ventricular interaction and relieves external constraint,^228,229 a mechanism that could work even in patients with a narrow QRS complex.²²⁹ The idea is that early LV contraction also allows early LV filling. In hearts with high central venous pressures, where presumably the RV occupies much of the pericardial space, the early LV activation may allow the LV to fill first, and thereby better, thus recruiting more of its Frank-Starling mechanism.²²⁸

A variety of cardiac and extracardiac processes triggered by CRT are responsible for its long-term beneficial effect. First, the improved pump function reduces neurohumoral activation, an effect evidenced by an increase in heart rate variability and a reduction in plasma brain natriuretic peptide (BNP) levels.²³⁰ Furthermore, the improved contractility and pump efficiency at a smaller end-diastolic volume reduce mechanical ventricular stretch. This latter reduction and the probably associated reduction in neurohumoral activation may well explain the beneficial reverse-remodeling effect of CRT.^225,231 As elegantly demonstrated by Yu et al.,²²⁵ such reverse remodeling points to structural improvement in the myocardium. These investigators showed the increasing reverse-remodeling effect of resynchronization over time by measuring the time course of LV cavity volume and LV dP/dt_max during the first 3 months after the start of CRT and 1 month after CRT was temporarily stopped, when some beneficial effect of CRT was still noticeable, indicating tissue adaptation (Fig. 9-30).

Figure 9-30 Acute hemodynamic and long-term reverse-remodeling effects of cardiac resynchronization therapy (CRT).

(Modified from Yu CM, Chau E, Sanderson JE, et al: Tissue Doppler echocardiographic evidence of reverse remodeling and improved synchronicity by simultaneously delaying regional contraction after biventricular pacing therapy in heart failure. Circulation 105:438-445, 2002.)

Study in canine LBBB hearts showed that isolated LBBB induces LV dilatation as well as hypertrophy and that these derangements are almost completely reversed by BiV pacing.^54,57 Interestingly, BiV pacing reversed the hypertrophy in the LV wall, especially in the late-activated LV lateral wall.⁵⁷ This reduction in hypertrophy is important because hypertrophy gives rise to various molecular changes,²³² resulting in greater risk for contractile failure and arrhythmias. Indeed, in dogs with LBBB, where heart failure was induced by 6 weeks of rapid atrial pacing, calcium homeostasis was affected most in the LV lateral wall.¹¹⁴ Notably, both hemodynamic improvement^36,185 and reverse remodeling⁵⁷ have been observed to a similar extent in dogs with LBBB (with and without heart failure) and in patients with primary and secondary dilated cardiomyopathies. Although CRT was originally indicated for patients with severe heart failure (NYHA Class III and IV), several studies demonstrated that CRT can also induce reverse remodeling in patients with only mild heart failure or impairment of LV function.^233,234 These data indicate that resynchronization cures “dyssynchronopathy” independent of the severity of heart failure.

Most CRT studies focus on the left ventricle, but a few indicate that RV function may also be improved by better synchronization of the electrical activation. This would especially apply to right bundle branch block (RBBB). Three-dimensional electroanatomic mapping data have provided the rationale for implementing CRT in these patients.²³⁵ However, rather than focusing on LV-based pacing, resynchronization of RBBB requires RV pacing. This also applies to children with repaired congenital defects such as tetralogy of Fallot, who present with right-sided heart failure.^236,237

Determination of Mechanical Asynchrony

The primary goal of CRT is to restore atrioventricular, interventricular, and intraventricular synchrony. Because QRS duration is a poor predictor of the response to CRT,^225,238,239 mechanical asynchrony was thought to be a better index for selection of CRT patients. Although studies have reported that measurement of mechanical asynchrony is a better predictor of the response to CRT than duration of the QRS complex, most were small and observational.

Interventricular asynchrony can be determined from the time difference between pulmonary and aortic valve openings with conventional echocardiography¹⁸⁶ or from the time to peak velocity in the RV and LV walls on tissue Doppler imaging (TDI). Other approaches are determination of the phase difference of RV and LV wall motion using nuclear imaging⁵³ and use of the time or phase difference with increased RV and LV pressures.^184,240 For practical reasons, the echocardiographic approach is most often used.

Intraventricular asynchrony usually requires more sophisticated equipment. The most detailed mechanical activation maps have been obtained through mapping of myocardial shortening (strain) patterns with magnetic resonance imaging (MRI) tagging^{48,201,209,241} (see Fig. 9-8). These studies have shown that the pattern of deformation gradually changes with increasing distance from the site of earliest activation (asterisk in Fig. 9-8). With the less expensive noninvasive method of M-mode echocardiography, the timing difference of maximal inward motion at the septum and posterior wall, or the septal-posterior wall motion delay (SPWMD),²³⁸ can be calculated. The disadvantages of this simple technique are that peaks of septal wall motion may not be precisely defined, and M-mode sections perpendicular to the septum may be difficult to acquire.

With color TDI and with pulsed-wave TDI, the time to peak systolic velocity can be measured in different regions of the myocardium.^225,242,243 Local contraction can be derived only through subtraction of velocities in nearby regions. Such analysis on digitally stored color-coded tissue Doppler images yields myocardial strain and strain rate.⁹⁷ Such information is equivalent to that obtained with MRI tagging, but with the great advantage that TDI can also be employed in the presence of a pacemaker. Strain measurements differentiate better than TDI between active systolic contraction and passive displacement. Likewise, Breithardt et al.⁹⁷ demonstrated more consistent behavior of strain than of velocity signals. However, image acquisition and analysis are time-consuming and operator dependent, and subtraction of two strain rate signals can create inaccuracies. Therefore, the clinical applicability of strain-rate imaging is limited.

A more recent development is the measurement of two-dimensional strain using speckle tracking. This technique uses the same principle as MRI tagging: measurement of the deformation of the tissue by displacement of markers in the tissue. Rather than markers artificially applied, as in MRI tagging, speckle tracking uses the differences in gray level naturally occurring in cardiac tissue as markers. Depending on the echocardiographic section used, longitudinal, circumferential, or radial strains can be measured. Although this technique shows promise and may have less technical limitations than TDI, the various echocardiographic systems appear to handle the strain signals differently in terms of tracking and spatial and temporal filtering, and the equipment allows only minimal interference by the user. Also, only a few studies showed a true validation of the technique, and this was not in conditions comparable to LBBB or CRT.

Nevertheless, speckle-tracking measurements have provided promising results. Mechanical dyssynchrony determined by this technique appeared to predict the response to CRT better than TDI,²⁴⁴ at least partly because strain measurement by speckle tracking is independent of the angle between ultrasound beam and the ventricular wall. TDI measures the velocity of displacement of the myocardial segments relative to the ultrasound transducer and in the direction of the ultrasound beam. Likewise, the longitudinal velocity, as usually determined in basal regions, is the sum of all velocities developed in all regions between the transducer and a particular area. Analysis of strain signals may also be less difficult to interpret than the frequently multiphasic velocity signals.²⁴⁵ TDI mechanical asynchrony is also observed in a considerable number of healthy volunteers.

Mechanical Asynchrony or Discoordination?

With mechanical asynchrony the focus over the last decade, the results of two randomized multicenter studies were unexpected. The PROSPECT and RethinQ studies showed poor predictive value for mechanical asynchrony as an index of CRT response in patients with wide and narrow QRS complex.^246,247 These disappointing results can be explained only partly by technical limitations of the TDI technique. Even when using the “gold standard” technique in cardiac deformation measurements, MRI tagging, determination of mechanical asynchrony (i.e., timing difference in onset or peak shortening) did not predict CRT response.^248,249

A better approach for achieving a mechanical predictor of CRT response may be to analyze discoordination of contraction. Whereas measurement of asynchrony requires determining the onset or peak of shortening in a certain region, discoordination is assessed by analyzing the strain curves over a large period of the cardiac cycle (e.g., systole or diastole). Reported indices include the CURE index,²⁴⁵ internal stretch fraction (ISF),²⁴⁹ and septal rebound stretch (SRS)⁵⁵ (Fig. 9-31). All three parameters provided a good prediction of CRT response, and better than using mechanical asynchrony measures in two studies.^55,249

Figure 9-31 Schematic representation of the determination of mechanical asynchrony (left panel) and discoordination (right panel). Mechanical asynchrony is often calculated from the time between the time to peak shortening in the septum (green) and lateral wall (red). Discoordination can be assessed as internal stretch fraction (ISF) or septal rebound stretch (SRS). AVC, aortic valve closure; AVO, Aortic valve opening.

The Site of Pacing

From a theoretical point of view, the “optimal” sites of pacing are those that establish the greatest reduction in total activation time. This requires that the slowly propagating electrical wavefront originating from the RBB is replaced by or fused with one or more activation wavefronts originating elsewhere. In current practice, CRT is most often achieved by pacing the RV and the LV simultaneously or sequentially. The RV pacing site used is usually the RV apex, and recent trials failed to show improved CRT efficacy during RV septal versus RV apical pacing.^250,251

The preferred LV pacing site is often the latest-activated region, which in LBBB is usually the basal part of the posterolateral wall.^39,40 By using extensive epicardial mapping and MRI strain analysis, Helm et al.²⁵² demonstrated that in dyssynchronous canine hearts, the area of late electrical and mechanical activation indeed closely matched the area that caused maximal LV dP/dt_max increase during pacing. This region spans about 40% of the LV free wall, especially the lateral wall. In similar studies, Rademakers et al.²⁵³ showed that in the presence of an infarction, the optimal site of LV pacing depends on the scar location. Gasparini et al.²⁵⁴ showed that placing the LV lead close to the ventricle’s basal portion conferred greater improvements in LV ejection fraction, quality of life, and distance walked in 6 minutes, similar to those reported for lateral lead placement. Their findings confirm previous observations by Butter et al.,²¹⁸ who showed that both anterior and lateral wall pacing sites improve LV contractility and pump function. However, pacing from the lateral wall resulted in a significantly larger increase in LV contractility; moreover, in about one third of patients, pacing from the anterior wall significantly depressed LV function. It is important to note that Butter paced the LV from a “middle-apical” position. High-resolution mapping data help explain these apparently discrepant findings (see Fig. 9-29). In patients with LBBB, the basal region of the heart is always the last region to be activated, with minor differences between the superior part of the left ventricle (around aortic root) and posterior and posterobasal regions;⁴⁰ the latter regions were the pacing sites selected by Gasparini.²⁵⁴ In contrast, the anterior middle-apical part of the left ventricle is usually early-activated, representing the LV breakthrough site, and activation of the middle-apical part of the lateral wall, one of the pacing sites tested by Butter et al.,²¹⁸ is significantly delayed.

Accounting for the differences in activation time in individual patients and using TDI to locate the region with the latest mechanical activation, Ansalone et al.²⁵⁵ investigated whether concordance of LV lead position and latest-activated region influences the outcome of CRT. Indeed, the greatest effect on functional parameters was seen when the LV lead was in the latest-activated region. Although optimization of the pacing site is often difficult when the transcoronary venous access is used, positioning the LV lead through limited thoracotomy may be helpful. During this procedure, LV pressure-volume analysis showed the importance of choosing the best position of the LV lead.²²⁶

During a coronary sinus approach, the left ventricle is paced at the epicardial surface, whereas under physiologic conditions, excitation of the LV initiates at the endocardium.¹¹ In a canine LBBB model, endocardial LV pacing during CRT consistently reduced electrical asynchrony compared with conventional epicardial CRT (Fig. 9-32). In parallel with these electrophysiologic improvements, endocardial CRT also improved systolic LV pump function and decreased dispersion of repolarization compared with epicardial LV pacing at the same site.³³ Additionally, the hemodynamic effects for endocardial sites were less dependent on location and AV delay than epicardial sites. LV endocardial pacing in patients can be established through a transatrial septal approach, shown to be technically feasible and effective during acute and midterm follow-up.^{206,256–259} However, its long-term effects compared to epicardial pacing and the development of systemic thromboembolic complications need to be investigated further.

Figure 9-32 Three-dimensional reconstruction of electrical activation times of the left and right ventricles during baseline LBBB (left) and biventricular (BiV) pacing with conventional epicardial (EPI) and novel endocardial (ENDO) left ventricular electrode position. Data from epicardial contact mapping and endocardial non-contact mapping in a canine heart with LBBB induced by radiofrequency ablation.⁵⁹

Given individual variations in etiology, severity, patterns of delayed ventricular activation, location of scar regions, and extent of mitral regurgitation in heart failure, it seems unlikely that one pacing site will “fit all.” This is further exemplified by studies in canine hearts with LBBB and with chronic myocardial infavetion, where the optimal site of LV pacing depended on the infarction site.²⁵³ Not surprisingly, pacing inside scar tissue, as in patients with a posterolateral scar, provides little benefit.²⁶⁰

Although stimulating the latest-activated regions makes sense, increasing evidence supports that pacing other areas may be at least as effective. In dogs, acute and chronic LV apical and LV septal pacing maintained regional cardiac mechanics, contractility, relaxation, and efficiency near native levels.⁵⁹ Derval et al.²⁵⁹ investigated 11 LV pacing sites (10 endocardial, 1 CS) in 35 nonischemic, dilated cardiomyopathy patients but failed to reveal a single optimal pacing site. However, selecting the best location individually doubled LV dP/dt_max compared with conventional coronary sinus pacing.

Given the electrical similarity of the spontaneously occurring LBBB and the LBBB induced by RV pacing, several centers have explored the feasibility of “upgrading” already-implanted RV pacemakers to BiV units or, in the case of a new implant, to use BiV pacing from the beginning. The results appear to depend on the clinical setting. In patients with already-implanted RV pacing leads, the upgrade leads to improvement in cardiac function and clinical outcome.^261,262 However, the situation is less clear when RV and BiV pacing are compared in patients with atrial fibrillation immediately after AV nodal ablation.^263,264 In these crossover studies, parameters such as quality of life, New York Heart Association (NYHA) diagnostic class, and 6-minute walking distance showed modest differences between RV pacing and BiV pacing. However, echocardiographic measures demonstrated moderate improvements with BiV over RV pacing.²⁶³ These findings are supported by experimental data showing that both RV pacing and BiV DDD pacing are able to completely reverse LV hypertrophy and dilatation induced by chronic AV nodal ablation in dogs.²⁶⁵ Together, these data suggest that the effects of ventricular pacing should be considered in light of the disease being treated.

Another important, yet unresolved issue is whether LV pacing alone is sufficient for adequate CRT. In dogs and patients with LBBB, LV pacing improves LV pump function at least as much as BiV pacing.^{36,161,219,220} LV pacing also leads to clinical improvement at up to 1 year of follow-up.^266,267 The promising effect of pacing at this single site can be explained by the fusion between the LV pacing–triggered wavefront and the wavefront created by intrinsic conduction through the RBB.³⁶ The Device Evaluation of CONTAK RENEWAL 2 and EASYTRAK 2: Assessment of Safety and Effectiveness in Heart Failure (DECREASE HF) study was a large, prospective randomized trial comparing single-site LV pacing to simultaneous and sequential BiV pacing.²⁶⁸ The three groups (each >100 patients) showed similar echocardiographic reverse remodeling after 3 and 6 months. Compared to LV pacing, simultaneous BiV pacing trended toward greater improvement in LV size. Importantly, however, the same AV delays were programmed for all groups, at approximately half the intrinsic PQ time. Therefore, during single-site LV pacing, the heart was predominantly depolarized by the activation wavefront originating from the LV electrode. Without electrical fusion with an activation wavefront from the RBB, the optimal effect of single-site LV pacing may be missed.³⁶

Even if LV pacing were sufficient for proper resynchronization, other factors would continue to warrant RV lead placement in many patients. In particular, current implantable defibrillator systems require an RV lead for tachyarrhythmia sensing and high-voltage therapies. Moreover, if a patient undergoing CRT also requires optimization of the AV interval because of prolonged PQ time or backup ventricular pacing, an RV lead would be important to allow pacing in case of dislodgement of the LV lead, which occurs in 5% to 10% of patients.

Atrial Pacing and Sensing

Given the rationale of CRT, continuous ventricular pacing should be provided to maximize pump function and cardiac efficiency. Many people equivocally equate a pacemaker with a resynchronizer, but even though they are remarkably similar in function and characteristics, the two devices fundamentally differ in one respect. The primary goal of a pacemaker is to restore the rhythm when absent; a secondary, less important, and even questionable goal (because of extreme difficulty in maintaining correct sequence of activation in ventricles during ventricular pacing) is to restore the proper AV timing in patients with first-degree AV block. In contrast, the primary goal of a resynchronizer is to resynchronize all the time, whether intrinsic activation is present or not. This difference creates opposite goals for the two devices in many situations.

A minimum heart rate is required to maintain adequate heart function without triggering heart failure. A minimum heart rate is also required to enable the heart to generate enough cardiac output to sustain the body’s metabolic needs. Nevertheless, we now know that cardiac efficiency is higher at lower heart rate (see Fig. 9-17) and that lower heart rates favor angiogenesis. A first attempt to define an adequate heart rate could be taken by stating that given these two opposing driving forces, the optimal heart rate for a given patient would be the lowest rate that creates sufficient cardiac output to satisfy basic metabolic needs in that patient at a given point. For a patient with normal ventricular function and no heart failure, a more aggressive stance could be taken with respect to programming higher parameters for lower rate limit (LRL) and rate response in the presence of sinus bradycardia or chronotropic incompetence. In the Dual Chamber and VVI Implantable Defibrillator (DAVID) trial, dual-chamber ICD pacing was programmed to an LRL of 70 bpm and short AV intervals; its results strongly suggest that these settings were worse than a simple setting of the ICD to VVI with an LRL of 40 bpm. Many now believe that most of the detrimental effects observed in the DDD arm of this trial resulted from RV pacing creating an asynchronous contraction.¹⁷⁵ In the DAVID-II study, VVI backup pacing at 40 bpm was compared with atrial pacing at 70 bpm. The data showed no significant difference between these two settings with respect to clinical outcome and survival as backup pacing.²⁶⁹ Therefore, a heart rate of 70 bpm does not seem to have a detrimental effect in the failing heart, as long as the sequence of conduction is normal.

To summarize, the data available strongly suggest caution in programming the pacemaker’s LRL to be higher than the sinus rate in patients with heart failure, whether or not they have sick sinus syndrome. This is especially important because RA pacing may cause interatrial conduction delays, leading to LA contraction against a closed mitral valve and thus lower cardiac output and elevated pulmonary pressure (Fig. 9-33). Moreover, higher-than-necessary heart rates worsen mechanical efficiency, decrease angiogenesis, exacerbate heart failure, and may even increase mortality (Fig. 9-34). In patients who receive an implant (CRT or CRT-D device) primarily for heart failure indications, a general rule is to program the LRL below the sinus rate, unless a clear clinical need to pace the atrium can be documented.

Figure 9-33 Left atrial (LA) contraction (contr) after mitral valve closure.

Top panel, Schematic illustration of a patient undergoing atrial pacing, in whom sensing in both ventricles has an interatrial conduction delay of 100 msec (ms) and, for the sake of simplicity, an equal mechanical delay. The patient’s intrinsic atrioventricular (AV) interval (marked as PR in the figure) is 150 msec. The figure illustrates how after a small electromechanical delay, the right and left atrial (RA and LA) contractions follow their electrical activations. This patient also has an intraventricular conduction defect, which makes the septum (marked as RV) activate 110 msec earlier than the lateral wall of the left ventricle (marked as LV). Similarly, both activations are followed after a small electromechanical delay by the septal and left lateral wall contractions. The mechanical asynchrony of the LV prevents the LA from contracting (contr) against a closed mitral valve. Bottom panel, Septum and left lateral wall are resynchronized with the optimum AV interval. Note clinically significant increase in filling time created by resynchronization. BV, Biventricular.

Figure 9-34 Relationship between the chronic changes in heart rate obtained with several drug therapies and their effect on mortality rates in patients with heart failure. Curiously, there is a strong correlation between mortality benefit and the ability of a drug to achieve a reduction in the heart rate of the patients in the active therapy arm of the trial. HDZ/ISDN, Hydralazine/isosorbate dinitrate; VHeFT, veteran Affairs Vasodilator Heart Failure Trial.

Atrioventricular Optimization

As discussed earlier, the AV interval strongly influences ventricular preload and thereby stroke volume in patients with reduced ejection fraction. During CRT, however, the AV interval also affects ventricular pump function by modifying ventricular asynchrony. At long AV intervals, the impulse conducted through the AV node and RBB is allowed to spread into the right ventricle and from there toward the left ventricle, maintaining (part of) the LBBB activation sequence. At shorter AV intervals, however, complete capture occurs from the pacing site(s), allowing resynchronization. The largest improvements in dP/dt_max occur at this point of transition from noncapture to complete capture. The best LV pump function is obtained with a relatively short AV interval, which allows complete activation of the ventricles from the two pacing sites.¹⁶¹ This AV interval may be different from the one resulting in optimized preload¹⁶⁴ (Fig. 9-35). In one half of patients, the AV interval for optimal preload is usually longer than that for optimal synchrony; thus it is no surprise that the mean difference between the two AV intervals is close to zero.¹⁶¹

Figure 9-35 Maximum preload is associated with maximum increase in pulse pressure.

Left, Upstroke of intraventricular pressure waveform in a patient with NYHA Class III heart failure (PATH-CHF-I trial). (a), During pacing at an intermediate AV interval; (b), during intrinsic propagation through the AV node of the normal atrial depolarization; (c), the ventricular electrogram during intrinsic activation; (d), the atrial electrogram. AVL, Time from peak of intraventricular pressure increase created by booster pump action of left atrium (A_P) and the start of the ventricular contraction (L_S); RA, right atrial. Right, Relationship between percent increase in aortic pulse pressure (PP) versus baseline (no pacing) obtained during cardiac resynchronization therapy (CRT) at different atrioventricular (AV) intervals (y-axis) versus the time between A_P and L_S for all the patients in whom hemodynamics improved with CRT in the PATH-CHF-I trial.

To date, prospective controlled data are still scanty to help determine the relative importance of optimizing synchrony and optimizing preload. Considering that asynchrony is a major cause of depressed pump function in patients with intraventricular conduction abnormalities, and that resynchronization has both short-term and long-term beneficial effects, we postulate that the primary target should be resynchronizing the ventricles, and that optimizing the ventricular inflow is of secondary importance in the majority of patients. The importance of restoration of ventricular asynchrony is supported by studies on the beneficial effect of CRT in patients with permanent atrial fibrillation (AF). In these patients, no preload optimization can be performed, and any possible beneficial effect on preload through CRT is excluded. Nevertheless, CRT has been shown to be effective in patients with AF and wide QRS complexes, especially once the AV node is ablated, abolishing any interference of underlying irregular, higher-rate AF spontaneous rhythm. The extent of improvement appears to be comparable to that seen in patients in sinus rhythm, in functional terms as well as long-term prognosis.^270–272

To optimize AV delay, invasive and noninvasive determinations of cardiac contractility and stroke volume may be performed. Invasive determination (e.g., using LV dP/dt_max)²⁷³ is probably more accurate but is cumbersome and expensive and may carry significant risk for the patient. Noninvasive measurements comprise diastolic inflow measurements (mitral valve E and A wave) and systolic outflow measurements (aortic velocity time integrals, finger arterial pulse pressure). The advantage of outflow measurements is accounting for possible influences of CRT on mitral regurgitation, but mitral inflow is usually easier to measure and more accurate.¹⁶⁹ Although validated to optimize AV interval in these patients, these echocardiographic methods remain time-consuming, with unproven clinical effect on large patient cohorts in prospective RCTs. Recent data suggest that the finger plethysmography for assessing arterial pressure and stroke volume is accurate²⁷⁴ and correlates well with echocardiographic methods for AV optimization in these patients.²⁷⁵

Integrated device–based algorithms to optimize AV interval have distinct clinical advantages, and different algorithms have been developed. One algorithm is based on measuring peak endocardial acceleration (PEA), a surrogate of LV dP/dt_max, using a lead with a microaccelerometer at the tip. The tip sensor detects vibration amplitude of the first heart sound, and this value should correlate with LV dP/dt.²⁷⁶ Other AV optimization algorithms are based on a combination of automatic measurements by intracardiac electrograms (EGMs) and surface ECGs²⁶⁸ or through indirect measurement of mechanical timing using surface ECGs or intracardiac signals.^277,278 Some studies have found excellent correlation between EGM-based algorithm and invasively derived dP/dt_max^268,279 or conventional echocardiography,²⁷⁷ whereas a more recent report demonstrated the reduced accuracy of a device-based interval algorithm optimization compared to conventional echocardiographic methods.²⁷⁸ Particularly relevant in this regard are the results of the large multicenter randomized trial SMART-AV that prospectively compared different modalities of AV interval optimization in CRT recipients, namely empirical AV delay of 120 msec, device electrogram-based algorithm, and echocardiographically guided AV optimization. Neither device electrogram nor echocardiographically based AV optimization conferred greater reduction of left ventricular end-systolic volume during follow-up compared to the empiric approach.^279a Other preliminary data were presented about the CLEAR and FREEDOM trials. The CLEAR trial investigated the effect of periodic AV and interventricular (VV) optimization using the PEA sensor. The clinical composite score had increased in 86% of periodically optimized patients versus 62% in the control group (P < .0007).²⁸⁰ Moreover, the proportion of patients who died from any cause or were hospitalized with heart failure was significantly lower (9%) in the sensor-based optimization group than the control group (25%; P < .001). In contrast, the FREEDOM study, using regular optimization with the Quickopt algorithm, did not improve CRT response.²⁸¹ Until now the evidence suggests that device-based algorithms for AV optimization may not be equivalent because each may lead to a different outcome.

Effect of Interventricular Delay

Most CRT devices now offer independent programming of AV and VV intervals. Studies on the effect of changing VV intervals during BiV pacing have reported conflicting results.^208,227,282 Because of variability in the conduction disturbances, as well as ischemic or scarred regions, different timing of stimulation of the two ventricles may be optimal.^283,284 In fact, more recent data from larger patient populations indicate the need for optimizing VV interval, with most patients requiring early LV offset. The largest prospective multicenter study evaluating the effectiveness of sequential pacing showed that sequential BiV pacing increased stroke volume and exercise capacity during follow-up to a greater extent than simultaneous BiV pacing for CRT.²⁸² Also, as with Vanderheyden et al.,²⁸⁵ this study confirms that about 60% of patients require early LV offset to optimize CRT. In the Rhythm-II ICD trial, patients with optimal programming of AV and VV interval showed better acute hemodynamic improvements; however, this did not translate into a difference in echocardiographic response after 6 months.²⁸⁶

An optimal setting of the VV interval may have a positive influence. According to the current model of how CRT works, the best resynchronization would be achieved by maximizing the synchrony of the activation through manipulation of the three activation wavefronts originating from the RBB, RV, and LV pacing leads. However, this same reasoning also indicates that an optimal VV interval may depend on the AV interval. The previously cited studies have not tested all the possible variations. Therefore, differences in optimal VV intervals may be caused, for example, by the use of different AV intervals. In clinical practice, performing such elaborate optimization protocol may be impractical and time-consuming, unless a simple technique such as finger blood pressure measurement is used,^275,284 with integrated optimization software that simultaneously elaborates and computes the noninvasive hemodynamic data. Optimal use of the extensive possibilities to vary timing of stimulation may be fully applicable only if medical equipment manufacturers provide adequate tools specifically designed and validated for the fine-tuning of AV and VV intervals to provide optimized CRT. It should be emphasized, however, that several prospectively designed studies evaluating the effect of VV timing optimization on symptoms and clinical outcome have failed to show a real clinical benefit. This might indicate that in the CRT population, VV timing plays a secondary role in modulating response to CRT. It is also possible, however, that a visible individual benefit may be diluted in the general population, or that the biologic variability offsets the minor (yet relevant in the individual patient) effect.