Intracardiac Echocardiography, Computed Cardiac Tomography, and Magnetic Resonance Imaging for Guiding Mapping and Ablation

Introduction

Noninvasive cardiac imaging with echocardiography, cardiac computed tomography (CCT), and/or cardiac magnetic resonance (CMR) imaging are increasingly available and used imaging modalities with which to define cardiac anatomy and function when planning and performing invasive cardiac electrophysiologic procedures ( Table 9.1 ). The technology underpinning these modalities is quite different, with these differences conferring particular advantages and limitations. Typically, echocardiography is performed in the first instance, with further testing guided according to a patient’s specific clinical situation and with consideration of potential adverse effects, contraindications, and availability.

Echocardiography

Echocardiography is the most widely available and established means of assessing cardiac structure and function. Piezoelectric crystals within echocardiographic transducers propagate and receive high-frequency sound waves. These ultrasound beams travel straight within homogeneous tissue but are reflected within heterogeneous tissue and at tissue interfaces. Motion (M-mode) and 2-dimensional echocardiographic images are generated in real time by interpreting these reflected ultrasound waves. Doppler echocardiography, which can be color-encoded, evaluates cardiac blood flow by using the ultrasound beam to detect changes in the frequency of the backscatter signal from moving red blood cells. In tissue Doppler imaging (TDI), lower velocity frequency shifts are analyzed to calculate the velocity of myocardial movement.

The reproducibility of volumetric measurements by echocardiography is inherently limited because estimates of volumes derived from 2-dimensional images are subject to variability and error imposed by reduced image quality, selection of the imaging plane, difficulties in identifying the endocardium-blood interface, geometric assumptions underlying the volumetric calculations, and beat-to-beat variations in volume and function. With the advent of 3-dimensional echocardiography, a more accurate and reproducible technique than 2-dimensional imaging, echocardiographic assessments of chamber size and cardiac function now correlate more closely with the gold standard, CMR.

The development of intracardiac echocardiography (ICE) was facilitated by technical advances in the construction of compact ultrasound transducers. The initial such ICE catheter was a mechanical ICE catheter, with a single ultrasound crystal mounted at the distal end of a catheter varying in diameter from 6 to 10 F. These catheters were typically nonsteerable, with the transducer connected to a motor in the handle of the device through a braided drive shaft. Engagement of the mechanism results in rapid 360-degrees rotation of the transducer, providing circumferential imaging in a place perpendicular to the long axis of the catheter. The 9 to 12 Hz ultrasound frequencies used by mechanical ICE catheters provides near-field clarity but poor tissue penetration and thus poor far-field resolution. These catheters have demonstrated utility in identification of endocardial anatomy, in the guidance of mapping and ablation of structures thus identified, in guidance of transseptal puncture, and in monitoring for procedural complications.

Phased-array ICE catheters were subsequently developed, incorporating a miniaturized 64-element, phased-array, electronically-controlled transducer mounted at the distal end of a steerable 8 or 10 F catheter. Imaging frequency can be varied from 5 to 10 MHz, the catheter tip can be deflected up to 160 degrees in two planes (anteroposterior and lateral), and the system permits the full range of 2-dimensional, M-mode, and Doppler imaging (including color, pulsed wave, continuous wave, and TDI). This catheter produces a wedge-shaped image (sector), which is displayed and manipulated on a conventional echocardiographic workstation identical to that used during transthoracic echocardiography (TTE) or transesophageal (TEE) echocardiography. Phased-array ICE has shown to provide image quality similar to that obtained with TEE, both when imaging the interatrial septum during device closure of an atrial septal defect and when imaging of the left atrial appendage (LAA) to allow Doppler interrogation of mechanical function in patients with atrial arrhythmias ( Figs. 9.1 and 9.2 ).

The use of ICE imaging has been shown to improve identification of endocardial structures and the accurate positioning of diagnostic and ablation electrophysiology catheters in relation to this defined anatomy, and this has markedly improved understanding of the relationship between anatomy and electrophysiology. In a series of seminal studies, ICE was used to define the tricuspid annulus, the crista terminalis, and the Eustachian ridge as important anatomic barriers to conduction and fundamental to the substrate supporting typical atrial flutter in humans. Subsequently Kalman et al., by using ICE imaging to accurately position a multipolar linear catheter along the crista terminalis, defined this structure as a major anatomic location and source focal atrial tachycardias that are readily amenable to catheter ablation. Accurate identification of the crista terminalis is also important in defining the location of the sinus node complex, and hence is important for guidance of sinus node modification procedures. Marchlinsky et al. have clearly demonstrated that, whilst the anatomic location of the crista terminalis is broadly understood, fluoroscopic identification of this structure is frequently inaccurate, and accuracy is greatly improved by the use of ICE imaging. The accurate demonstration of cardiac anatomy with ICE is particularly important in the setting of complex cardiac anatomy, such as patients with surgically-repaired congenital heart disease. ICE has been demonstrated to allow precise definition of anatomic structures, to monitor adequate catheter-tissue contact, and to facilitate safe atrial baffle puncture.

ICE has been demonstrated to establish and hence improve the stability of contact between an ablation catheter tip and a target tissue surface, leading to more efficient ablation. Ablation catheters can be readily visualized by ICE imaging and have a typical appearance with a bright tip and a fan-shaped acoustic shadow. Kalman et al. have compared the traditional criteria for determining tissue contact (stable electrograms and fluoroscopic images) with ICE-guided ablation in a canine model. Without ICE guidance, lateral sliding of the ablation catheter (>5 mm) was frequent (18%), as was poor perpendicular electrode-tissue contact (27%), leading to smaller lesion formation and a lower efficiency of heating index (i.e., the ratio of steady-state temperature to power). For linear ablation arrays, ICE improves the accuracy of positioning and the extent of tissue contact in animal models compared with fluoroscopy alone. The additional value of ICE when using of contact force sensing catheters has not been formally studied.

ICE imaging allows the real-time assessment of lesion formation and size. In an excised canine heart model of ablation, Kalman et al. used high-frequency ICE imaging (15 MHz) to measure lesion size after application of radiofrequency (RF) ablation and demonstrated a high correlation between ultrasonic and pathologic depth. Although other investigators have used ICE to observe lesion formation in vivo, it is less clear that a correlation exists between the ablation-induced changes observed in a beating heart, such as swelling and increased echogenicity, and lesion size.

In a series of human and swine studies, Marchlinsky et al. used ICE to describe the nature of lesion formation as a result of the application of RF energy. They observed the development of mural swelling and increased tissue echodensity after the application of RF in the right atrium (RA). Furthermore, after transmural linear-lesion formation in the posterior RA of pigs, the demonstration of mural swelling by ICE correlated with the finding of mural edema on histopathologic analysis. In a study of cavotricuspid isthmus (CTI) ablation by Morton et al. using phased-array ICE and an imaging frequency up to 10 MHz, it was possible to observe discrete lesion formation after RF application, which manifested predominantly as regions of tissue swelling. Interestingly, without the application of further RF energy, ICE demonstrated a progression over ensuing minutes to a more diffuse swelling and lesion coalescence, likely representing development of tissue edema after RF application.

This swelling at the location of RF application has had important clinical implications in some anatomic locations. In humans, ICE has been used to demonstrate narrowing of the junction between the RA and the superior vena cava (SVC) during RF application at the superior aspect of the crista terminalis during sinus node modification procedures. Indeed, SVC syndrome has been described as a complication of circumferential RF ablation at this junction, an observation that predated by several years the observation of pulmonary vein (PV) narrowing as a result of circumferential ostial PV ablation.

Cardiac Computed Tomography

CCT is an imaging technique that, as with traditional X-ray imaging, used ionizing radiation to rapidly generate multiple cross-sectional grey-scale images of the body. Traditionally, the most common clinical indication for CCT is the assessment of coronary arteries for the presence of luminal plaque, and accurate localization of the epicardial coronary arteries remains of vital importance in many invasive electrophysiologic procedures, but CCT can also accurately measure cardiac chamber dimensions and define both cardiac anatomy and function. Multidetector row scanners acquire between 64 and 320 slices (0.6-mm slice thickness) of cardiac tissue within a single rotation of the gantry on which the instrument is mounted. The timing of image acquisition following intravenous administration of iodinated contrast is determined to enable optimal opacification of the cardiac structure of interest. Image quality can be impaired with heart rates greater than 60 beats per minute and by irregular heart rhythms such as atrial fibrillation (AF). As mentioned, CCT is associated with a small dose of radiation exposure and may be contraindicated in patients with allergies to iodinated contrast or significant renal impairment.

CCT is ideally suited to providing anatomic information regarding cardiac chamber geometry, and such anatomic images may readily be integrated into 3-dimensional mapping platforms during atrial or ventricular mapping and ablation procedures. Other anatomic information of relevance to mapping and ablation that has been made available by contemporary contrast-enhanced CCT techniques with submillimeter spatial resolution and acquisition times below 0.4 seconds includes static and dynamic measures of wall thickness, with such measures able to predict the locations of abnormal endocardial bipolar voltage in the left ventricle (LV) during electroanatomic mapping, for example. In addition, the presence of scar and physiologic variables such as myocardial perfusion may be assessed and quantified, again with areas of CCT-demonstrated LV myocardial hypoperfusion correlating well with areas of reduced bipolar electrogram voltage on invasive mapping. Tian et al. have demonstrated that, in addition to a simple endocardial surface reconstruction, CCT images demonstrating abnormal anatomic and physiologic characteristics of ventricular myocardium can be extracted and integrated into clinical 3-dimensional mapping systems.

Cardiac Magnetic Resonance Imaging

Over recent years, CMR has evolved in both clinical and research settings to provide gold standard volumetric assessments of chamber structure and function, and to provide detailed characterization of cardiac tissue. The high spatial and temporal resolution capabilities of CMR provide several advantages over other imaging modalities. Furthermore, contrast-enhanced CMR with gadolinium-based agents has revolutionized the noninvasive assessment of cardiac fibrosis. However, CMR remains relatively expensive, is limited in its availability in many healthcare systems, and patient-specific factors such as claustrophobia can limit its utility. The presence of implanted cardiac devices or metallic foreign bodies has also previously been regarded as a contraindication, although this has changed markedly with the development of magnetic resonance imaging (MRI)-conditional devices and with greater clinical experience. In patients with significant renal dysfunction, administration of gadolinium contrast may lead to nephrogenic systemic fibrosis.

CMR images are generated using the magnetization attributes of cardiac tissue. During CMR scanning, hydrogen protons are tilted off the longitudinal (z) axis of the scanner into the transverse plane (x and y axes) by applied pulse sequences. Following restoration of longitudinal magnetization of the protons, tissue-specific T ₁ and T ₂ relaxation properties can be determined. The T ₁ relaxation time reflects the time decay constant for 63% recovery of the longitudinal magnetization equilibrium value of a proton, whereas T ₂ relaxation time represents the decay of the transverse magnetization signal to 37% of its original value. Different cardiac tissues exhibit varying relaxation times depending on the molecular environment of water molecules, and these qualities are used to construct pixel-based images. T ₁ and T ₂ properties are also influenced by pathologic processes such as fibrosis and inflammation.

Gadolinium-based contrast agents are inert and cannot move across intact cell membranes. Following intravenous gadolinium administration, contrast enters the extracellular space, shortening the tissue’s T ₁ relaxation time. Passage of contrast into and out of the extracellular space is influenced by tissue perfusion, the extracellular volume of distribution, and the specific kinetic profile of the contrast agent. Cardiac fibrosis prolongs the “wash-out” period of gadolinium because of a reduction in the tissue’s capillary network. Regions of cardiac tissue in which there is increased accumulation of gadolinium contrast appear as bright signal intensity on T ₁ -weighted sequences, referred to as late gadolinium enhancement (LGE). Such imaging relies upon identifying a difference in signal intensity between fibrotic and “normal” myocardium. This means its utility in evaluating more diffuse patterns of interstitial cardiac fibrosis as compared with confluent replacement fibrosis is limited.

Diffusely fibrotic cardiac tissue, with diffuse expansion of the extracellular matrix between cardiomyocytes, accumulates contrast in a similar manner to regional scar, but calculation of the global T ₁ time is necessary for its quantification. Such quantification requires the acquisition of multiple images to derive the T ₁ recovery curve. T ₁ time is inversely proportional to the concentration of contrast within the extracellular space, therefore more cardiac fibrosis leads to lower T ₁ times. Several “T ₁ mapping” techniques have been correlated with myocardial collagen content in humans, but potential confounding factors have been described, including the dose of contrast, glomerular function, body composition, hematocrit, and acquisition time after the contrast bolus.

Role of Cardiac Imaging in Mapping and Ablation of Atrial Fibrillation

Preprocedure Imaging

Left atrial (LA) size has proven to be a powerful predictor of many outcomes in management of AF. In particular, LA size as a marker of the degree of underlying remodeling predicts the likelihood of maintaining sinus rhythm after electrical cardioversion or catheter ablation for AF. Nedios et al., using CCT imaging, demonstrated a gradient from paroxysmal to longstanding persistent AF in LA volume and in asymmetry of chamber geometry, and demonstrated the magnitude of change in both parameters to predict AF recurrence after ostial PV isolation. Similarly von Bary et al., using CCT or CMR to measure LA volume and TTE to measure LA diameter and area, demonstrated all three parameters to be significant predictors of recurrent AF, and LA volume to predict recurrence with a persistent rather than a paroxysmal phenotype. In a systematic review of 22 studies relating LA diameter measured through M-mode imaging of the LA to the risk of AF recurrence after a first PV isolation, Zhuang et al. reported LA diameter to be a key predictor of procedural success, with between-study heterogeneity in results chiefly governed by the duration of follow-up and the intensity of surveillance for asymptomatic AF recurrences.

The extent of atrial remodeling may, of course, be ascertained by a myriad of imaging modalities and geometric variables beyond simple determination of LA size. Kurotobi et al. have used CCT to demonstrate that, as the LA enlarges, the shape of the LA roof initially becomes flat and then becomes coved, and this progression correlates with less physiologic reliance on PV triggers and a higher incidence of inducible atrial tachyarrhythmias even after complete PV isolation. Pursuing a similar concept, Bisbal et al. used contrast-enhanced CMR to demonstrate that a more spherical rather than a more discoid LA is associated with greater LA volume, a persistent AF phenotype and more structural heart disease, and is an independent risk factor for AF recurrence after a single catheter ablation procedure.

Risk stratification through any of these techniques relies on accurate cardiac imaging. The most commonly available modality remains TTE, which has the advantages of imposing no radiation exposure and of providing a wide array of clinically useful data, but the disadvantages of being relatively time consuming and reliant on highly skilled operators. Measurement of LA size by TTE has long been standardized. Volumetric analyses are preferred to linear measurements of LA diameter, and 3-dimensional TTE imaging, which provides direct identification of the LA endocardial border in the collected dataset, is preferred to 2-dimensional imaging, which relies on area measurements to derive LA volumes. Volumetric data generated by CMR and CCT imaging appear comparable to data from 3-dimensional TTE.

Increasingly multislice CCT and CMR, which can provide both static LA volumes and dynamic volumes that allow assessment of phasic atrial function, are used for evaluation of LA size. CCT does, of course, expose patients to ionizing radiation, and use of CMR has been limited in the setting of implanted cardiac devices but, because of relatively poor image quality in TTE, the interobserver variability in both CMR (3 ± 10%) and CCT (1 ± 11%) estimation is lower than with 2-dimensional TTE (9 ± 24%). Furthermore both LA volumes and functional measures have been shown to correlate closely between CCT and CMR, despite the somewhat lower spatial resolution of CMR. LA volumes appear to be overestimated by approximately 10% by CCT when the examination is performed in AF, but the correlation between the values thereby derived remains between 0.83 and 0.85.

The structural remodeling manifest by LA enlargement is believed to be fundamentally driven by myocardial fibrosis. A number of imaging techniques have been developed to quantify LA fibrosis, as a guide to the potential success of catheter ablation. One such technique is LGE CMR, which relies on gadolinium becoming trapped within the expanded extracellular matrix of myocardial regions containing replacement fibrosis and which therefore would be expected to be washed out more slowly than it would be from nonfibrotic myocardial regions. On T ₁ -weighted imaging the result is increased signal intensity. As mentioned, it is a qualitative technique that requires the presence of regions of normal myocardium as a frame of reference to generate image contrast.

In their initial report, Oakes et al. reported the extent of LA delayed enhancement by LGE CMR to correlate strongly with the extent of low endocardial bipolar voltage during invasive electroanatomic mapping (R ² = .61; P <.05), and the rate of recurrent AF over 10 months was strongly associated with the extent of LA enhancement (14% for minimal enhancement; 75% for extensive enhancement). A systematic scoring system for the extent of delayed enhancement (Utah I-IV; Fig. 9.3 ) has subsequently been tested as a predictor of the outcome of AF ablation in a prospective multicenter study. In 329 enrolled patients, it was possible to quantify tissue fibrosis in 83% at a single core laboratory. Two hundred and sixty participants with paroxysmal (65%) or persistent AF were then followed for 475 days after a first catheter ablation procedure. The risk of recurrent AF increased from 15.3% for stage I fibrosis (< 10% of the atrial wall) to 69.4% for stage IV fibrosis (≥ 30% of the atrial wall), and the addition of fibrosis to a multivariate model that included variables recognized to predict recurrence markedly improved the predictive accuracy of the model. Various authorities have, however, highlighted the need to further improve methods of segmenting the LA image, to improve hyperenhancement thresholds to accurately identify replacement fibrosis, and to improve reproducibility of data analysis before LGE CMR can be considered a routine clinical tool.

Another CMR-based technique that has been explored is postcontrast T ₁ mapping of the LA, which allows direct signal quantification. After a direct relationship between T ₁ relaxation time and the extent of ventricular interstitial fibrosis was demonstrated by Iles et al., Beinart et al. applied postcontrast T ₁ mapping to the posterior wall of the LA and demonstrated significantly shorter times in patients with AF compared with control patients without AF. Adjusting for age, AF phenotype, hypertension, and diabetes, longer relaxation times were reported to be significantly associated with higher LA endocardial bipolar voltage. Finally, Ling et al. used a T ₁ mapping sequence to determine the postcontrast T ₁ relaxation time at the interatrial septum as an index of diffuse LA fibrosis in 20 control patients, 71 patients with paroxysmal AF, and 41 with persistent AF. Again, a positive relationship was reported between T ₁ relaxation time and bipolar voltage (R ² =.48; P <.001), and a long T ₁ relaxation time was found to be the only independent predictor of AF recurrence 12 months after first catheter ablation of AF. There does, however, continue to be debate regarding the reproducibility of this technique.

Simple volumetric analysis of LA function can give an insight into the potential efficacy of catheter ablation. The LA regulates LV filling, acting as a reservoir during ventricular systole and then as a passive conduit from the PVs to the LV during early diastole. In late diastole, specifically during sinus rhythm, it acts as a booster pump for LV filling. Phasic function of the LA can be assessed by volumetric analysis during the cardiac cycle, which can be based on echocardiographic, CCT, or CMR imaging. Maximum (before mitral valve opening), minimum (during mitral valve closure), and precontraction volumes allow total, passive, and active emptying fractions to be determined, respectively reflecting reservoir, conduit, and booster pump functions. Dodson et al., for example, used volumetric CMR analysis to relate the LA passive emptying fraction, as an index of conduit function, to the risk of AF recurrence after PV isolation. Over more than 2 years, 124 of 346 patients experienced recurrent AF, and a strong independent association was found between the passive emptying fraction and the risk of recurrence (hazard ratio for lowest vs. highest quintile = 3.92; 95% confidence interval [CI], 2.01–7.65).

Another method of studying phasic atrial function involves measurement of myocardial deformation. Initially, this was performed through TDI techniques applied sequentially to multiple LA myocardial segments to determine strain and strain rate, delineated as strain per unit time. Again, measures of conduit, reservoir, and contractile LA function can be determined. A more contemporary adaptation of strain imaging, which avoids the angle-dependence of Doppler techniques and is more reproducible, is 2-dimensional speckle-tracking echocardiography. This involves tracking of characteristic acoustic speckles throughout the cardiac cycle, with deformation of all atrial segments seen in the apical views quantified simultaneously. Global longitudinal strain and strain rate values can be determined in 94% of study subjects, with good reproducibility. Using this technology, strain and strain rate values indicative of LA reservoir and conduit functions have been found to be reduced in the setting of diabetes and hypertension, even before the development of more advanced atrial remodeling manifest by LA enlargement. In at-risk adults, an LA strain measure reflecting reservoir function has been demonstrated to be an independent predictor of incident cardiovascular events, superior to traditional risk factors including LA area and indexed LA volume. In patients with nonvalvular AF, an LA strain measure reflecting poor reservoir function has been shown to be a more powerful predictor of thromboembolic risk than a traditional algorithm based on clinical risk factors, and in chronic AF, there is a demonstrated inverse relationship between stroke risk and LA reservoir function, independent of age and LA volume.

Multiple investigators have directly tested variables of LA function derived using 2-dimensional speckle-tracking echocardiography as noninvasive markers of LA myocardial fibrosis. In both patients undergoing surgical repair for severe mitral regurgitation because of mitral valve prolapse in the absence of AF and in a cohort with either mitral stenosis or regurgitation of whom 64% were in AF at the time of imaging, a close correlation has been observed between low LA strain and more advanced fibrosis. This relationship appears stronger than that between fibrosis and indexed LA volume, LA ejection fraction, or the E/E’ ratio.

Several groups have compared the results of echo-based strain imaging with the extent of fibrosis as assessed by the Utah staging system. In 55 patients with paroxysmal or persistent AF, including 29% in AF at the time of the relevant studies, significantly lower strain values were reported in patients with persistent compared with paroxysmal AF, and lower strain in the lateral LA wall was an independent predictor of a greater extend of LA fibrosis on LGE CMR imaging (r = -.5; P =.006). More recently, Watanabe et al. compared LA strain imaging with the extent of myocardial fibrosis on invasive electroanatomic maps. Using the heterogeneity in time to peak strain between 16 LA segments as the primary outcome variable in 52 patients with paroxysmal AF, heterogeneity was significantly higher in patients with regions of low bipolar voltage (14.1 ± 5.7 vs. 8.0 ± 5.1; P =.0002). A significant association was also reported between increased heterogeneity and a longer LA activation time (r = .57; P = .0001), further suggesting that strain imaging can provide a meaningful noninvasive index of the extent of LA electrical and structural remodeling.