Cardiac MRI in Identifying Ventricular Arrhythmia Substrate
Cardiac MRI in Identifying Ventricular Arrhythmia Substrate
Patients with a documented VA in a 12-lead surface ECG submitted for ablation and who underwent ce-CMR before the index procedure were included in the study. Patients without 12-lead ECG of the clinical VA [arrhythmia recorded by the implantable cardioverter defibrillator (ICD) or observed in a 24-h Holter] and an induced VA in the index procedure matching the cycle length of the clinical VA were also included. Patients were considered for VA ablation if they met one of the following criteria: (i) an incessant VT, (ii) repetitive episodes of sustained monomorphic VT, or (iii) symptomatic frequent premature ventricular contractions (PVCs) despite the use of antiarrhythmic drugs, associated with structural heart disease (SHD). Patients with claustrophobia or classical contraindication for a ce-CMR acquisition were excluded. Patients with idiopathic VA not associated with SHD or with the diagnosis of arrhythmogenic right ventricular dysplasia (ARVD) were also excluded. All participants signed their informed consent, and the study protocol was approved by the local Ethics Committee.
Patient-identifying data were removed from ce-CMR images for the analysis, performed by two-independent cardiologists also blinded to any clinical and electrophysiological data. In a case of discordance, a third observer was required. The ce-CMR images were analysed to depict the presence of scarred tissue for each left ventricle (LV) segment according to the 17-segment model. The right ventricle (RV) septum was divided into three segments: right ventricular outflow tract (RVOT), right ventricular inflow tract (RVIT), and right ventricular apex. Equivalence to the 17-segments model of the RV septum was as follows: RVOT—Segment 2; RV apex—Segments 8, 9, and 14; RVIT—Segment 3. The RV free wall was not taken into account for the analysis, because it was too thin in the spatial resolution of the ce-CMR images to reliably establish a degree of hyper-enhancement (HE) transmurality. For details of the ce-CMR sequence, see Supplementary Material Online.
Accordingly, the HE pattern distribution for each segment was defined as follows:
Endocardial ce-CMR pattern: HE involved the endocardium with a mean transmural extent ≤50% of wall thickness.
Transmural ce-CMR pattern: HE involved the endocardium with a mean transmural extent >50% of wall thickness.
Epicardial ce-CMR pattern: HE involved the epicardial layer with a mean transmural extent <50% of wall thickness.
Mid-myocardial ce-CMR pattern: HE extended to <50% of wall thickness with mid-myocardial distribution.
Absence: No HE was identified in the corresponding segment.
Figure 1 shows examples for each of the five HE types. Sensitivity and specificity analyses were performed both for the presence of one or more segments with epicardial HE and for a predominance of epicardial HE in the ce-CMR images.
(Enlarge Image)
Figure 1.
Pattern distribution of hyper-enhancement in cardiac magnetic resonance images. (A) Endocardial hyper-enhancement. In this case, Segments 13, 14, and 16 (red arrow) presented endocardial hyper-enhancement. (B) Transmural hyper-enhancement. The contrast-enhanced cardiac magnetic resonance of this patient showed a transmural hyper-enhancement in Segment 4 (red arrow) and Segment 5. Segment 3 was also affected by endocardial hyper-enhancement. (C) Mid-myocardial hyper-enhancement. In this case, mid-myocardial hyper-enhancement affects Segments 2 (red arrow) and 3. Segment 4 is partially affected. (D) Epicardial hyper-enhancement. Red arrows show epicardial hyper-enhancement in Segments 10 and 11. In these images, it is also possible to observe mid-myocardial hyper-enhancement in Segments 7 and 8.
All participants underwent an electrophysiology study in a fasting state, with oral sedation (10 mg diazepam). Intravenous conscious sedation was used during the electrophysiology study, except in cases of PVC ablation. In PVC cases, ventricular mapping was performed without intravenous sedation and a bolus of fentanyl was administered intravenously before radiofrequency (RF) ablation. A navigation system (CARTO system, Biosense Webster, Diamond Bar, CA, USA) was used to guide the VA ablation. A tetrapolar diagnostic 6-Fr catheter was introduced through the femoral vein and placed at the RV apex. The basal and post-ablation stimulation protocol consisted of programmed ventricular stimulation from the RV apex at three drive cycle lengths with up to three extrastimuli and incremental burst pacing at a cycle length up to 200 ms. If the clinical VT was not inducible, intravenous isoproterenol was used. A 3.5 mm electrode irrigated-tip catheter (Thermocool Navistar®, Biosense Webster) was introduced through transseptal or retrograde aortic access for LV endocardial mapping. A non-surgical transthoracic epicardial access was performed for epicardial mapping and ablation when endocardial VT ablation was unsuccessful, when the endocardial mapping did not identify a VT substrate, when the patient had a non-ischaemic cardiomyopathy and the ECG was suggestive of an epicardial origin (before endocardial mapping), or when ce-CMR showed an epicardial scar.
As scar is usually present in more than one segment and could have different distribution patterns in each one of them, the ventricular segment related to the VA was estimated on the basis of ECG morphology of VT using the Miller classical algorithm. The 17-segment model used for the image analysis and location of the VA origin was adapted to the classical electrophysiology region model (Figure 2).
(Enlarge Image)
Figure 2.
Adaptation of the electrophysiological 12-segment model to the 17-segment model. Adaptation of the 17-segment model to the anatomical regions described by Miller et al. As a limitation of this algorithm, an important number of regions were not identified as a possible ventricular arrhythmia site of origin.
However, the Miller algorithm applies only to ischaemic patients and there is no validated algorithm to identify the origin of VA in non-ischaemic patients. Therefore, the sensitivity and specificity analyses in predicting an epicardial origin of the VA were performed only for the successful ablation segment.
Radiofrequency delivery was temperature-controlled with a power limit of 50 W at a target temperature of 45°C at the endocardium. At the epicardium, the ablation settings were 40 W/45°C, and occasionally the power was increased to 50 W. The flow rate was 0 and 17 mL/min during mapping and RF ablation, respectively. Ablation of sustained VT was guided by the identification of diastolic electrograms and entrainment mapping criteria. In the case of unmappable VT, pacemapping manoeuvres identified the exit site in the scar and a substrate mapping and ablation approach was performed. In 39 (48.8%) patients, a substrate ablation was performed in addition to the clinical VT ablation. The same manoeuvres were used for both endocardial and epicardial ablation of the clinical VA. The endpoint of the ablation procedure was the suppression of clinical VA inducibility and of any monomorphic VT, whatever the approach needed. The actual SOO of the clinical VA was determined by the location of the successful ablation site for that VA and was recorded in the CARTO electroanatomical map.
The follow-up was performed every 6 months after the ablation. The visit included clinical evaluation, a 24-h Holter, and an echocardiogram. The definition of PVC recurrence was >5% of PVC burden in repeated 24-h Holter monitoring (1, 6, and 12 months), and of VT recurrence as any VT episode requiring ICD intervention or documented by any means.
Patients with mid-myocardial HE in the successful ablation site were analysed in more detail. The distance was measured between the border/centre of the HE and the endo-/epicardium in the case of an LV free wall or between the border/centre of the HE and the endocardial RV and LV surface in a case of a septal scar. This measurement refers to the thickness of the healthy myocardium interposed between the scar and the tip of the ablation catheter at the target ablation site. These patients were excluded from the general sensitivity and specificity analyses of the scar distribution pattern to predict the epicardial origin.
Quantitative variables are expressed as mean value ± SD, and qualitative variables are expressed as the number and percentage. The sensitivity and specificity, as well as the positive predictive value (PPV) and negative predictive value (NPV), were obtained for the following: (i) the presence of any segment with epicardial HE, (ii) a majority of segments with epicardial HE, and (iii) the type of HE in the successful ablation segment. Comparisons were made using the two-sided χ test or the Wilcoxon test as appropriate. Statistical significance was set at P < 0.05. All data were analysed using the PASW Statistics 18.0 software (SPSS, Inc., Chicago, IL, USA).
Methods
Patient Sample
Patients with a documented VA in a 12-lead surface ECG submitted for ablation and who underwent ce-CMR before the index procedure were included in the study. Patients without 12-lead ECG of the clinical VA [arrhythmia recorded by the implantable cardioverter defibrillator (ICD) or observed in a 24-h Holter] and an induced VA in the index procedure matching the cycle length of the clinical VA were also included. Patients were considered for VA ablation if they met one of the following criteria: (i) an incessant VT, (ii) repetitive episodes of sustained monomorphic VT, or (iii) symptomatic frequent premature ventricular contractions (PVCs) despite the use of antiarrhythmic drugs, associated with structural heart disease (SHD). Patients with claustrophobia or classical contraindication for a ce-CMR acquisition were excluded. Patients with idiopathic VA not associated with SHD or with the diagnosis of arrhythmogenic right ventricular dysplasia (ARVD) were also excluded. All participants signed their informed consent, and the study protocol was approved by the local Ethics Committee.
Contrast-enhanced Cardiac Magnetic Resonance Analysis
Patient-identifying data were removed from ce-CMR images for the analysis, performed by two-independent cardiologists also blinded to any clinical and electrophysiological data. In a case of discordance, a third observer was required. The ce-CMR images were analysed to depict the presence of scarred tissue for each left ventricle (LV) segment according to the 17-segment model. The right ventricle (RV) septum was divided into three segments: right ventricular outflow tract (RVOT), right ventricular inflow tract (RVIT), and right ventricular apex. Equivalence to the 17-segments model of the RV septum was as follows: RVOT—Segment 2; RV apex—Segments 8, 9, and 14; RVIT—Segment 3. The RV free wall was not taken into account for the analysis, because it was too thin in the spatial resolution of the ce-CMR images to reliably establish a degree of hyper-enhancement (HE) transmurality. For details of the ce-CMR sequence, see Supplementary Material Online.
Accordingly, the HE pattern distribution for each segment was defined as follows:
Endocardial ce-CMR pattern: HE involved the endocardium with a mean transmural extent ≤50% of wall thickness.
Transmural ce-CMR pattern: HE involved the endocardium with a mean transmural extent >50% of wall thickness.
Epicardial ce-CMR pattern: HE involved the epicardial layer with a mean transmural extent <50% of wall thickness.
Mid-myocardial ce-CMR pattern: HE extended to <50% of wall thickness with mid-myocardial distribution.
Absence: No HE was identified in the corresponding segment.
Figure 1 shows examples for each of the five HE types. Sensitivity and specificity analyses were performed both for the presence of one or more segments with epicardial HE and for a predominance of epicardial HE in the ce-CMR images.
(Enlarge Image)
Figure 1.
Pattern distribution of hyper-enhancement in cardiac magnetic resonance images. (A) Endocardial hyper-enhancement. In this case, Segments 13, 14, and 16 (red arrow) presented endocardial hyper-enhancement. (B) Transmural hyper-enhancement. The contrast-enhanced cardiac magnetic resonance of this patient showed a transmural hyper-enhancement in Segment 4 (red arrow) and Segment 5. Segment 3 was also affected by endocardial hyper-enhancement. (C) Mid-myocardial hyper-enhancement. In this case, mid-myocardial hyper-enhancement affects Segments 2 (red arrow) and 3. Segment 4 is partially affected. (D) Epicardial hyper-enhancement. Red arrows show epicardial hyper-enhancement in Segments 10 and 11. In these images, it is also possible to observe mid-myocardial hyper-enhancement in Segments 7 and 8.
Electrophysiological Study
All participants underwent an electrophysiology study in a fasting state, with oral sedation (10 mg diazepam). Intravenous conscious sedation was used during the electrophysiology study, except in cases of PVC ablation. In PVC cases, ventricular mapping was performed without intravenous sedation and a bolus of fentanyl was administered intravenously before radiofrequency (RF) ablation. A navigation system (CARTO system, Biosense Webster, Diamond Bar, CA, USA) was used to guide the VA ablation. A tetrapolar diagnostic 6-Fr catheter was introduced through the femoral vein and placed at the RV apex. The basal and post-ablation stimulation protocol consisted of programmed ventricular stimulation from the RV apex at three drive cycle lengths with up to three extrastimuli and incremental burst pacing at a cycle length up to 200 ms. If the clinical VT was not inducible, intravenous isoproterenol was used. A 3.5 mm electrode irrigated-tip catheter (Thermocool Navistar®, Biosense Webster) was introduced through transseptal or retrograde aortic access for LV endocardial mapping. A non-surgical transthoracic epicardial access was performed for epicardial mapping and ablation when endocardial VT ablation was unsuccessful, when the endocardial mapping did not identify a VT substrate, when the patient had a non-ischaemic cardiomyopathy and the ECG was suggestive of an epicardial origin (before endocardial mapping), or when ce-CMR showed an epicardial scar.
ECG-suggested Ventricular Segment of Origin
As scar is usually present in more than one segment and could have different distribution patterns in each one of them, the ventricular segment related to the VA was estimated on the basis of ECG morphology of VT using the Miller classical algorithm. The 17-segment model used for the image analysis and location of the VA origin was adapted to the classical electrophysiology region model (Figure 2).
(Enlarge Image)
Figure 2.
Adaptation of the electrophysiological 12-segment model to the 17-segment model. Adaptation of the 17-segment model to the anatomical regions described by Miller et al. As a limitation of this algorithm, an important number of regions were not identified as a possible ventricular arrhythmia site of origin.
However, the Miller algorithm applies only to ischaemic patients and there is no validated algorithm to identify the origin of VA in non-ischaemic patients. Therefore, the sensitivity and specificity analyses in predicting an epicardial origin of the VA were performed only for the successful ablation segment.
Radiofrequency Ablation
Radiofrequency delivery was temperature-controlled with a power limit of 50 W at a target temperature of 45°C at the endocardium. At the epicardium, the ablation settings were 40 W/45°C, and occasionally the power was increased to 50 W. The flow rate was 0 and 17 mL/min during mapping and RF ablation, respectively. Ablation of sustained VT was guided by the identification of diastolic electrograms and entrainment mapping criteria. In the case of unmappable VT, pacemapping manoeuvres identified the exit site in the scar and a substrate mapping and ablation approach was performed. In 39 (48.8%) patients, a substrate ablation was performed in addition to the clinical VT ablation. The same manoeuvres were used for both endocardial and epicardial ablation of the clinical VA. The endpoint of the ablation procedure was the suppression of clinical VA inducibility and of any monomorphic VT, whatever the approach needed. The actual SOO of the clinical VA was determined by the location of the successful ablation site for that VA and was recorded in the CARTO electroanatomical map.
The follow-up was performed every 6 months after the ablation. The visit included clinical evaluation, a 24-h Holter, and an echocardiogram. The definition of PVC recurrence was >5% of PVC burden in repeated 24-h Holter monitoring (1, 6, and 12 months), and of VT recurrence as any VT episode requiring ICD intervention or documented by any means.
Mid-myocardial Hyper-enhancement Analysis
Patients with mid-myocardial HE in the successful ablation site were analysed in more detail. The distance was measured between the border/centre of the HE and the endo-/epicardium in the case of an LV free wall or between the border/centre of the HE and the endocardial RV and LV surface in a case of a septal scar. This measurement refers to the thickness of the healthy myocardium interposed between the scar and the tip of the ablation catheter at the target ablation site. These patients were excluded from the general sensitivity and specificity analyses of the scar distribution pattern to predict the epicardial origin.
Statistical Analysis
Quantitative variables are expressed as mean value ± SD, and qualitative variables are expressed as the number and percentage. The sensitivity and specificity, as well as the positive predictive value (PPV) and negative predictive value (NPV), were obtained for the following: (i) the presence of any segment with epicardial HE, (ii) a majority of segments with epicardial HE, and (iii) the type of HE in the successful ablation segment. Comparisons were made using the two-sided χ test or the Wilcoxon test as appropriate. Statistical significance was set at P < 0.05. All data were analysed using the PASW Statistics 18.0 software (SPSS, Inc., Chicago, IL, USA).