Study design
We selected all consecutive patients scheduled for stand-alone thoracoscopic surgical ablation of AF in the Academic Medical Center (AMC), Amsterdam, between January and December 2013. Patients were identified from a prospectively entered registry. We performed a cross-sectional comparison of TTE and CE-MRA exams made during standard work-up for AF ablation surgery and followed patients for one year to determine success of the procedure. The Institutional Review Board waived written informed consent. The study was conducted in accordance with the Declaration of Helsinki.
Study population
Patients undergoing thoracoscopic surgery in our centre have advanced AF, defined as persistent AF according to the ESC 2016 guidelines for the management of AF [1], enlarged left atria [6] or one or more previously failed catheter ablation14. Patients must have failed at least one class Ic or class III anti-arrhythmic drugs.
Procedure and follow-up
The standard thoracoscopic surgical AF ablation procedure of the AMC was described in detail previously [14, 15]. All patients underwent radio frequent ablation of the left and right PV antrum (AtriCure Isolator™ Synergy™ bipolar RF ablation clamp). In persistent AF patients, additionally a superior line and trigone line were created (AtriCure Isolator™ Transpolar™pen). All ablation lines were tested for bidirectional block with epicardial electrodes connected to an EP system [16]. The four main ganglionated plexus were ablated, unless patients participated in the AFACT trial and were randomized to no GP ablation [14].
After AF ablation surgery, patients were followed in accordance with the 2012 HRS guidelines, which included a visit to the outpatient clinic every 3 months with ECG and Holter monitoring to determine AF recurrence in the first year. After an initial 3-month blanking period, all antiarrhythmic drugs (AADs) were discontinued. AF recurrence was defined according to current guidelines as a 30 s continuous rhythm registration of AF, or an ECG recording of an AF episode in patients not using AAD [5].
Transthoracic echocardiography
All patients underwent 2D triggered TTE (Vivid 9, GE VingmedUltrasound AS, Horten, Norway). All TTE examinations were made specifically to determine LA volume and function during the work-up for AF ablation. Four- and two-chamber views were obtained by experienced cardiac echocardiographists according to the recommendations of the American Society of Echocardiography [6]. Recordings were made using a 1.6-MHz to 3.2-MHz transducer (System 9; GE Healthcare, Milwaukee, WI), digitized, and analysed offline with EchoPAC (GE Healtcare, 2015 General Electronic Co.).
Maximum LA volume was defined as the tracing of the LA endocardial borders at the largest visual volume which is one or two frames prior to the opening of the mitral valve in the apical four- and two-chamber views and was calculated using the modified method of discs. In patients with AF at the time of TTE, we visually selected the frame with the largest volume from all available frames to eliminate the effect of reduced cardiac filling at short R-R intervals. We excluded the left atrial appendage (LAA), PVs and area between the mitral valve annulus and mitral valve leaflets from the LA volume determination (Fig. 1a) [6].
Contrast enhanced magnetic resonance angiography
CE-MRA was performed using a 1.5 MRI scanner (Somatom Avanto MR B17 Siemens, Erlangen, Germany) with a 6 element body matrix surface coil. A standard dose of gadobutrol (0.1 mmol/kg of Gadovist; Bayer Vital GmbH, Leverkusen Germany) was injected by a power injector and through a 20-gauge plastic cannula placed in an antecubital vein for all examinations. For the MRI acquisition, a standard automated bolus injection of 0.1 mmol/kg body weight of gadobutrol (Gadovist, Bayer Schering Pharma, Germany) was used at a flow rate of 2 cc/sec, followed by 20 cc of saline flush, at the same rate. The MRI sequence with centric ordering of k-space was started manually as soon as the contrast agent was seen in the left atrium on the2D real-time fluoroscopy. We used a routine breath-hold technique and 2D contrast bolus tracking sequence to time the first pass CE-MRA acquisition. Imaging parameters were: 96 slices, FOV 500 × 344 mm, image resolution 1.3 × 1.3 × 1.3 mm, parallel imaging factor 2, image acquisition time 19 s.
LA volumes and dimensions were analysed offline using image analysis software (Medis Mass & Flow, Leiden, the Netherlands, version 2015-EXP). The LA was segmented using the CE-MRA images and LA volume was assessed by manual tracing of the endocardial borders of successive slides in the sagittal plane (Fig. 1b, c). The atrioventricular groove and mitral valve were used as landmarks to separate the LA from the left ventricle. The LAA and PVs were excluded, analogous to LA volume assessment with TTE. As an indication for LA enlargement in one direction, the maximal craniocaudal, anteroposterior and transversal dimensions were measured in the maximal sagittal and transversal planes respectively.
Repeated measurements
All patients underwent a CE-MRA and 2D TTE prior to AF ablation surgery. Two independent readers (N.v.d.B. and D.C.P.Y) retrospectively analysed all stored CE-MRA and TTE images to measure LA volume. A random sample of 25% was selected for repeated measurements by one observer to evaluate the intra-observer variability.
Statistical analysis
Data are presented as frequencies, mean with SD or median with interquartiles as appropriate. TTE and CE-MRA LA volumes were defined as the mean of the volumes measured by two observers for each modality. Differences between LA volume on TTE and CE-MRA were evaluated using the Pearson correlation coefficient and intraclass correlation coefficients (ICC) with 95% confidence intervals (CI) as described by Shrout and Fleiss [17]. Consequently Bland-Altman analyses were constructed to demonstrate bias in units (ml) and as a percentage of the mean volume (%, bias/average of TTE and CE-MRA volume*100) and to assess the limits of agreement (LoA), defined as 1.96SD around the mean [18]. A Pearson correlation between the mean volume and volume difference was applied to determine the presence of proportional difference variability. Additionally, the Bland-Altman plots were used to identify the 10 most extreme outliers between TTE and CE-MRA volume measurements. The stored images of the most extreme outliers were retrospectively assessed and discussed for cues to explain the discrepancies. We determined the ICC of each modality for inter-observer and intra-observer variability. The analyses of inter-modality, inter-observer and intra-observer variability were performed for the entire study population and subsequently for stratified subgroups of patients known to be congruously in sinus rhythm or AF during both TTE and CE-MRA acquisitions.
Receiver-operator curve (ROC) analysis was performed with calculation of the area under the curve (AUC). The sensitivity and specificity of the cut-off value based on the Youden’s index was provided. ROC analyses were used to determine the discriminative value of LA volumes and CE-MRA dimensions to determine AF recurrence after thoracoscopic surgery.
Analyses of volume measurements were performed with both non-indexed volume measurements and volume measurements indexed for BSA. Because of evident similarities between the non-indexed and indexed measurements, we here reported the non-indexed measurements for technical comparisons and indexed values where relevant. For clinical comparisons, such as ROC analysis of outcome, the indexed volumes were used. All performed tests were two-sided and values of P < 0.05 were considered statistically significant (IBM SPSS statistics version 24 and R, version 3.2.3).