Since the initial description that in the majority of patients paroxysmal atrial fibrillation (AF) is triggered by pulmonary vein (PV) ectopy,1 catheter ablation of AF has gone from a niche procedure to a common one, with approximately 20,000 ablations performed per year in the US alone. However, AF ablation remains a technically difficult procedure, requires long training to become proficient,2 and success rates are still only in the magnitude of 70–80% for a single procedure for patients with paroxysmal AF. Additionally, the left atrium is anatomically complex and is adjacent to multiple structures that can easily be damaged during the ablation, with catastrophic consequences.3–9
Until the last decade the only procedural imaging options available to the electrophysiologist were fluoroscopy (see Figure 1), with transoesophageal and transthoracic echocardiography being of occasional and limited use. Although this gave the physician an accurate depiction of the catheters, the heart, being soft tissue, was not well visualised. This led to the development of electroanatomic mapping systems whereby the ablation catheter could be localised in 3D space by the use of magnetic fields or changes in impedance (see Figure 1). The major limitation of these systems was that although a 3D geometry could be built up over time, the operator has only limited information regarding tissue contact. This, and the complex anatomy of the left atrium, led to rather crude maps being made.
The left atrium and, importantly, the anatomy of the PVs are highly variable among the population. Although four PVs are commonly present (left superior PV, left inferior PV, right superior PV and right inferior PV),10 variations are common,11 with common ostia for ipsilateral veins being reported in up to 83% on the left and 40% for right-sided PVs. Additional veins are also common, for example a right middle vein has been reported in up to 22% of patients,11 and further locations of PVs, although rare, have been demonstrated to be able to perpetuate AF; for example veins connecting directly to the roof of the left atrium.12 For this reason an accurate understanding of each individual’s left atrium and PV connections is essential for a successful procedure.
Initially the first advanced imaging modality to be used successfully was computed tomography (CT). A pre-procedural left atrial CT gave the clinician valuable information before operating. However, one of the limitations was that the CT was not displayed concurrently with the fluoroscopic imaging that the electrophysiologist used to perform the procedure. To overcome this obvious limitation, systems were developed to overlay the segmented left atrium from the CT directly onto the live fluoroscopy screen.13 For the first time this allowed the operator to have a 3D representation of the patient’s cardiac anatomy displayed over realtime fluoroscopy. Although the segmented left atrium had to be manually registered to the patient, various techniques were tested and were found to be accurate to within 0.3mm. A benefit of the CT overlay was that, in addition to the cardiac anatomy, extra-cardiac structures could also be displayed, for instance the oesophagus.
Although CT overlay was an improvement, there were some important limitations: the patient has to undergo the scan prior to the procedure, which normally occurs some days before; the position of the patient in the CT scanner is often different to how the patient is during the AF ablation, and the volume status of the patient is also likely to be different. These differences can lead to changes in the left atrium; however, whether these are of clinical relevance is unknown.
Rotational angiography was designed to overcome some of these latter drawbacks of overlaid CT.12,14–18 With rotational angiography, the left atrium is opacified with contrast, with users choosing to either inject the contrast in the inferior vena cava, the right atrium, the right ventricular outflow tract, or the left atrium. Once opacified, the C-arm of the X-ray fluoroscopy system rapidly rotates in a 240° arc around the patient. This data set is then segmented in the same way as if a pre-procedural CT had been acquired. The benefits of rotational angiography are that it is performed during the procedure, with the same X-ray equipment as is used for fluroscopy, and, furthermore, fluid status and patient position are the same as displayed on the live fluoroscopy screen. Additionally, there are no concerns regarding registration of the segmented anatomy onto the live fluoroscopy, as this is accomplished automatically, assuming the patient does not move. If the patient does move during the procedure the anatomy can be quickly re-registered by using the trachea as an anatomical landmark. In contrast with the 3D electroanatomical mapping systems that have been in use for over a decade, if the rotation is performed prior to the transseptal puncture, when using a right-sided injection protocol (we use the right atrial inferior vena cava junction), segmentation and registration of the image can be performed concomitantly. In a recent study comparing rotational angiography with a 3D electroanatomical mapping system without integration of CT, no differences were seen in procedural outcome, procedural time or radiation exposure.19 The time to make the map again was similar for both groups, but the additional time for the rotational angiography was significantly shorter as the segmentation can be completed before entering into the left atrium, whereas the 3D electroanatomical map has to be made once access has been gained to the left atrium.
Although rotational angiography was a step forward in terms of accurately imaging the left atrium, compared with stand-alone 3D electroanatomical systems, the obvious advantage that these systems have is the integration of electrophysiological data. A recent development with rotational angiograpy is the integration of data taken from the electrophysiology (EP) recording system (see Figure 2). Although in prototype form at present, some data have been presented that demonstrate proof of concept, with the ability to perform activation maps, dominant frequency analysis and voltage mapping. How this integrated approach compares with the latest 3D electroanatomical mapping systems has not yet been investigated. So far, these imaging tools give the electrophysiologist information that could be ascertained by just using simple fluoroscopy and considerable operator experience. Whether these systems actually lead to improved patient outcomes is uncertain. In one retrospective study there was a suggestion that 3D electroanatomical systems with integrated CT imaging of the left atrium led to better procedural outcomes.20 However, a later prospective randomised trial by the same group did not demonstrate any difference in patient outcomes when CT integration was used or not.21 The conclusions that were drawn from these studies were that pulmonary vein isolation is paramount, however the procedure is performed, and that there is a learning curve effect when using CT integration. When operators are able to see the complex anatomy of the left atrium with CT, this greater understanding of the anatomy leads to better outcomes even when CT integration is not used. A recent retrospective study that demonstrated improved clinical outcomes when using a 3D electroanatomical system with live integration of CT did show that when CT integration was used, more PVs were identified and targeted compared with when CT integration was not used. This underlines the importance of accurately understanding each patient’s anatomy, rather than relying on the misconception that patients have just four PVs.22
The next generation of imaging tools may offer new information to the electrophysiologist. Currently, tissue contact is not routinely assessed, although this is known to be a key determinant of lesion development.23 A number of different systems are being developed so that tissue contact can be assessed and controlled, ranging from robotic24–27 or magnetic navigation of the catheter27–30 to incorporation of a force tip sensor in the catheter tip.31,32 Although this is a step in the right direction, currently all forms of lesion assessment are indirect, ranging from change in tissue impedance, diminution of local electrogram voltage, and contact force assessment. Magnetic resonance imaging (MRI) offers the ability to assess scar formation in the left atrium. Although MRI of the left atrium is still in its infancy a number of different groups have shown interesting data regarding radiofrequency (RF) ablation of AF.33,34 In a case report, a gap in a linear lesion could be seen on pre-procedural MRI and this correlated with a gap in the line that was responsible for an atrial tachycardia.35 Ablation at this gap was able to terminate the arrhythmia and result in electrophysiological block of the line. A further study has looked at the correlation between points where energy has been delivered, and marked using a 3D electroanatomical system and scar formation as assessed by MRI.36 The results of this study demonstrated that there was a relatively poor correlation between the two, with 20% of sites that energy was thought to have been delivered to resulting in no scar formation as assessed by late gadolinium enhancement.36 A different study has shown that oedema formation as assessed by T2-weighted imaging may contribute to acute PV isolation, with the concern that when such oedema resolves PV reconduction could occur.37
Although MRI is able to give us information regarding scar formation, currently this is only performed following the procedure (see Figure 3). What the electrophysiologist requires is to know this information in realtime. Realtime ultrasound offers this possibility, and work has been presented that demonstrates that an ultrasound transducer integrated into an RF ablation catheter is accurately able to detect changes in tissue structure that correlate with tissue necrosis.38 Although this work has so far been performed only in animals, it is hoped that clinical trials are forthcoming.
In summary, we have gone from using fluoroscopy, which gives relatively little information regarding cardiac anatomy, to having a number of imaging modalities to choose from. Although all of these options have their relative merits, there are few data that conclusively show improved patient outcomes. This is perhaps due to trial design. There is obviously a considerable learning curve to AF ablation that may be accelerated by using the advanced imaging modalities discussed. A limitation of AF ablation is that, currently, we have no way to visualise lesion formation, which is, after all, key to a successful outcome. Future imaging solutions are being developed that give information that is currently unavailable, namely visualisation of pre-existing scar, scar formation and oedema formation in response to ablation. Hopefully this additional information will result in clinical benefits to patients.