The first and most extensively studied ion channel implicated in the development of short-term CM was I_to. In an acute open-chest canine model Del Balzo et al. used 4-aminopyridine (4-AP), which blocks I_to blocker to abolish the T wave changes induced by 20-min ventricular pacing [26]. Based on the preferential expression of I_to in canine epicardium [46] the mechanism of short-term CM was proposed to include a change in transmural voltage repolarization gradient. Further support for this hypothesis was found by studying action potentials in isolated endocardial and epicardial slabs while simulating induction of CM by pacing in the direction perpendicular to the fiber direction. This intervention abolished phase 1 notch, prolonged the action potential of the epicardial cells, and reduced the transmural repolarization gradient. Pretreatment of the slabs with 4-AP reduced the phase 1 notch in the epicardial tissue and no further changes in action potential shape and duration were observed during pacing [47].

In long-term CM experiments prolongation of the epicardial action potential duration [25], loss of the phase 1 notch, decrease in I_to density, and decreased mRNA levels for Kv4.3, the gene controlling the pore-forming unit of the channel carrying canine I_to [31] and in the protein of its accessory subunit, KChIP2 [29] were observed in the left ventricular epicardium, indicating a role for I_to in the evolution of long-term CM as well (Fig. 25.2).

While the direction of changes in I_to density and distribution is similar in short- and long-term CM, their mechanisms appear to be different. In short-term CM the decrease in I_to is believed to result from internalization of the AT1 angiotensin receptor/Kv4.3/KchIP2 complex (discussed further below) [48]. In long-term CM it is achieved by the reduction in the transcription factor CREB (cAMP response element binding protein). Critical to this observation is a cyclic AMP response element near the promoter region of KChIP2 and Kv4.3. A reduction in CREB would lead to decreased synthesis of the channel components. Both the short- and long-term processes appear to be initiated by angiotensin II synthesis and release and may involve the actions of reactive oxygen species [29].

Several additional ion channels are implicated in short- and long-term CM including I_Kr [32], I_Ca,L [24], possibly by the mechanism of CREB affecting transcription of some of them [29]. The reversal of the normal epicardial > endocardial gradient for the I_Kr density and the loss of the gradient for ERG mRNA and protein were documented in long-term CM around the ventricular pacing site also contributing to the local epicardial APD prolongation [32]. Reduction in epicardial connexin43 expression close to the pacing site (but not at remote ventricular areas) with gap junction redistribution was also observed along with the decrease in the conduction velocity around the pacing site [49] and an increase in the activation time during normal conduction [33] in long-term CM. A schematic of the short- and long-term CM development is presented in Fig. 25.3.

It is important to point out that the majority of the biochemical studies in CM involved tissue samples obtained from the areas 1–2 cm away from the left ventricular pacing site and although a gradient change in the transcription pathways across the ventricle was documented [34] the regional differences (especially in relation to the local mechanical function) have not been fully explored.

The processes of ventricular remodeling operate in the different directions in early- and late-contracting areas. The late results of this differential remodeling of the ventricle are quite opposite: thinning and atrophy of the early-­contracting areas and hypertrophy of the late-contracting regions [44, 45]. Therefore it is plausible to predict that different mechanisms of remodeling operate in different areas of the ventricle at earlier stages as well which requires further investigation.

Along with the signal transduction and molecular control of CM its triggering factors have been extensively studied. It was clear that changes on a cellular and molecular level cannot be directly explained by the change in electrical activation sequence itself (a single cell cannot possess the knowledge of the timing of the activation front); therefore other mechanisms have to be involved. From the early experiments the alteration in the local stress-strain relationship and geometry of ventricular contraction triggering activation of the renin-angiotensin system was suspected as the CM trigger and confirmed experimentally. Pretreatment of animals with saralasin, (AT1 and AT2 receptor antagonist), captopril (angiotensin converting enzyme inhibitor), and chymostatin (inhibitor of tissue chymase, which converts angiotensin I to angiotensin II) abolished the development of short-term CM suggesting a role for local angiotensin II production in its initiation [50].

Additional proof of local angiotensin II release as the CM trigger came from the demonstration that AT1 receptors co-localize with the α (alpha) (Kv4.3) and accessory (KChIP2) subunits of the I_to channel. Exposure to angiotensin II caused internalization of the AT1R–Kv4.3-KChIP2 macromolecular complex [48] reducing I_to current in HEK cells in which the channel components of the complex were overexpressed. The results were validated in co-immunoprecipitation experiments in cardiac myocytes, and in experiments using FRET. Another mechanism by which altered strain triggers CM is the stretch activated receptors (SAR) [2, 40]. In a canine model SAR blockade with streptomycin dramatically attenuated CM [40].

Sosunov et al. [51] in an isolated perfused rabbit heart obtained direct evidence that it is indeed mechanical factors and not the electrical activation sequence itself that serves as the primary CM trigger. The excitation-contraction uncoupler blebbistatin completely abolished CM after ventricular pacing, while local myocardial stretch without ventricular pacing (and therefore unchanged activation) produced T wave changes similar to CM. This observation suggests potential common mechanisms of T wave changes observed in CM and various forms of structural heart disease associated with mechanical dyssynchrony without significant QRS prolongation such as myocardial infarction, hypertrophy, certain cardiomyopathies, etc.

While the studies of molecular mechanisms associated with CM in a large part have reached consensus, the electrophysiological effects of CM at the level of the whole heart are less clear.

There is an ongoing debate regarding to what degree transmural vs regional repolarization gradients are responsible for the T wave changes in CM. In perfused myocardial wedge preparations Jeyaraj et al. found APD prolongation in late- (and, to lesser extent early-activated regions) suggesting that regional repolarization gradient is responsible for long-term CM, while small transmural repolarization gradient remained unchanged [1]. In contrast, Spragg et al. using ablation-induced LBBB demonstrated shortening of action potential duration, effective refractory periods, and conduction velocity in the wedge preparations from the late-activated areas with increased tissue strain [2]. APD shortening in the late-activated areas was also consistently observed in the isolated rabbit heart [19, 51]. In the intact in-situ canine heart Coronel et al. showed the development of the transmural repolarization gradient localized at the area close to the pacing site with no evidence for significant regional gradient during long-term CM [33]. In-situ study of the short-term CM by the same group showed baseline apical-basal repolarization gradient that did not change significantly although repolarization time shortened uniformly in the setting of short-term CM [30].

Using non-invasive ECG imaging (ECGI) in patients with WPW syndrome before and after ablation of an accessory pathway Ghosh et al. detected the presence of a large (85–95 ms) regional repolarization gradient between the accessory pathway ventricular insertion site and the rest of the myocardium (translating into apical-basal gradient given the annular location of the accessory pathways). This gradient was caused by a significantly prolonged activation-recovery time in the area of the bypass tract insertion that gradually resolved within 1 month consistent with the time course of the long-term CM [52].

The challenges observed in extrapolating the results obtained in isolated myocytes, tissue slabs, and wedge preparations to the effects of CM in the heart in situ mirror those in explaining the genesis of the normal T wave [53, 54] (see [55] for potential explanations and discussion).

The relevance of these findings to human CM (the major contributors to which are right ventricular pacing and LBBB) is unclear as well. The vast majority of animal CM models followed the original idea of complete reversal of left ventricular activation by stimulating the latest activated site during normal conduction (i.e. posterolateral LV base) [36, 37]. However the degree of mechanical dyssynchrony which (rather than the electrical conduction delay) is the primary CM trigger does not correspond to the degree of activation reversal. Instead, endocardial right ventricular apical (rather than the epicardial left ventricular basal) pacing was found to create the most dyssynchronous contraction [39, 43]. This is one of the possible reasons why CM following right ventricular pacing in humans achieves the steady state more rapidly (1 week) [56, 57] than left ventricular pacing.

Since the introduction of the original concept of cardiac memory in 1982 and until recently, two important issues remained unresolved.

First, it was universally accepted that the return to the normal ventricular activation sequence is pre-requisite to observe it. Although it was clear that CM develops gradually during the period of abnormal activation secondary T wave changes were thought to completely obscure it.

The most common causes of CM in clinical practice – LBBB and atrioventricular block requiring ventricular pacing are rarely reversible. This significantly limited clinical application of CM as the majority of the patients developing CM never return to normal conduction.

Second, although obvious individual variations in the degree of CM, were observed [70], it was unclear what factors determine this variability. The only factor associated with CM magnitude was the amount of ventricular pacing [57].

Using further refinement of the quantitative methods of CM assessment developed previously in the animal studies [27], we sought to resolve these issues.

In our clinical practice we observed that following pacemaker implantation the amplitude of T waves during ventricular pacing decreases in time. We hypothesized that this change in T wave amplitude is related to the development of CM. Using vectorcardiographic software [71] we compared 3 dimensional T vector loop displacement after 1 week of ventricular pacing (DDD with short atrioventricular delay) in humans in AAI mode (conventional measure of CM) to that in DDD mode with continuous ventricular pacing [58].

We found that not only the 3-dimensional direction of the T wave shift was identical when measured in AAI and DDD mode (Fig. 25.5), but the magnitude of the shift in the two pacing modes correlated closely (Fig. 25.6). In contrast to AAI mode (conventional CM) which demonstrated typical rotation of the T wave vector towards the paced QRS direction, CM in DDD mode (continuous ventricular pacing) was characterized by a progressive decrease in discordant T vector magnitude without change in direction. The strong correlation between T vector changes in both pacing modes suggested that they reflect the same physiological phenomenon. These findings have important implications.