Clinical experience regarding safety and diagnostic value of cardiovascular magnetic resonance in patients with a subcutaneous implanted cardioverter/defibrillator (S-ICD) at 1.5 T

Between August 2017 and June 2019, 23 consecutive patients with an implanted S-ICD (EMBLEM, Boston Scientific, Marlborough, Massachusetts, USA; MR-conditional) were admitted to our CMR Centre for further diagnostic work-up of suspected ischemic/non-ischemic cardiomyopathy, myocarditis and/or ventricular arrhythmia of unknown origin by multi-parametric CMR imaging. To ensure that only patients without second active or abandoned implanted devices, components or accessories such as lead adaptors, extenders or leads were included, pre- and post-implantation chest x-rays and complete patient history were validated. Furthermore, there was no evidence of fractured electrodes or compromised pulse generator system integrity. All patients gave their written informed consent prior to the CMR study.

The CMR study was performed in supine position on a 1.5 T CMR scanner (Ingenia, Philips Healthcare, Best, the Netherlands) in breath-hold and with electrocardiogram (ECG)-triggering with a maximum specific absorption rate (SAR) ≤ 2 W/kg (sequence parameters, Table 1). Since prominent hyperintensity artefacts in the localizer images may prevent appropriate planning of standard imaging axis, single shot (black-blood gradient-echo) acquisitions that are less prone to off-resonance artefacts could be used for planning in case of need. Balanced steady-state free precession (bSSFP)- and/or spoiled gradient echo (GRE)-based cine-CMR images were acquired in short-axis (SAx) and three long-axis (LAx) views. In addition, T2-weighted (T2w) short tau inversion-recovery (STIR) edema imaging was performed in three SAx slices. Aortic flow measurements were performed at the level of the aortic valve and/or ascending aorta using velocity-encoded cine (VENC)-CMR. Thereafter, first-pass myocardial perfusion imaging (MPI) was performed in three SAx slices (basal, mid-ventricular and apical) at resting conditions (0.075 mmol/kg body weight gadolinium (gadobutrol) was injected with a rate of 3 ml/s intravenously followed by 30 ml saline flush at the same rate) using a saturation-recovery bSSFP sequence accelerated with sensitivity encoding. Then, 3D-CMR angiography of the thoracic aorta was performed (with another infusion of 0.1 mmol/kg gadobutrol). Finally, late gadolinium enhancement (LGE)-images (T1-weighted inversion recovery gradient-echo sequence) were acquired 10–15 min after the second intravenous contrast administration in the same imaging planes as the cine-images. In case of substantial hyperintensity artefacts, alternative wide-band inversion-recovery LGE (wb-LGE) sequences were applied. Moreover, optimization of images with reduction of image artefacts was individually achieved by either acquiring images with breath-hold in inspiration, by using non-standard views and/or by changing the phase-encoding direction. VENC-CMR for aortic flow measurement, resting MPI and 3D-CMR angiography of the aorta were performed as part of our routine comprehensive CMR protocol, since an additional infusion of gadobutrol is not required if LGE-imaging is an essential part of the protocol.

Before starting with the CMR study, the S-ICD was programmed to the MR protection mode. Hence, the VT/ventricular fibrillation (VF) therapy mode of the device was deactivated during the CMR study. Patient monitoring during the CMR examination comprised continuous ECG-, respiratory-, pulse oximetry- and blood pressure-monitoring. Following the CMR study, the device interrogation and reprogramming was repeated.

CMR data analysis was performed off-line by two experienced readers using commercially available software (CVi42, Circle cardiovascular imaging, Calgary, Alberta, Canada). Ventricular volumes, ejection fraction and LV mass were derived by contouring endo- and epicardial borders on the short-axis cine images and indexed to body surface area (BSA). Manual delineation of the aortic lumen was performed throughout the cardiac cycle in order to obtain aortic flow values. T2w-edema, resting MPI, 3D-CMR angiography of the aorta and LGE images were visually assessed.

Assessment of image artefacts was performed visually for each CMR sequence using the following four categories: 1 = no image artefact at all and no limitation in image interpretation; 2 = good image quality with limited artefacts (affecting less than 25% of the ventricles or the target structure); 3 = poor image quality with substantial artefacts (affecting more than 25% but less than 75% of the ventricles or the target structure), and 4 = meaningful image assessment impossible due to extensive artefacts (affecting more than 75% of the ventricles or the target structure). In addition, a more detailed assessment of image artefacts was performed for cine- and LGE-imaging using a standard 17-segment model.

As illustrated in Table 2, the mean age of our study population was 48 ± 12 years (range 32–65 years) and 43% (N = 10) were women. Indications for S-ICD implantation comprised suspected ischemic (4%) and non-ischemic cardiomyopathy (22%) with a left ventricular (LV) ejection fraction (LVEF) ≤35%, suspected myocarditis and/or ventricular arrhythmias of unknown origin (74%). In the majority of our patients (65%), S-ICD implantation was performed for secondary prophylactic reasons.

A successful CMR study could be performed in all patients without any case of unexpected early termination being required due to technical and/or vital complications (Table 3). No power reset, battery discharge, reprogramming, neurological symptoms or thermal sensation were observed. Lead impedance and function were unaffected in all patients after the scan. The only technical issue that was observed during this study, was a complete loss of the S-ICD beeper volume in 52% of subjects after the CMR study (after having contact with the magnetic field). During the CMR study, premature ventricular complexes were monitored in 35% of our patients – without any need to change the study protocol.

The detailed results regarding assessment of CMR image quality and artefact extent for each CMR sequence used in the present study are given in Table 4. Notably, there was no relevant artefact caused by the subcutaneous electrode of the S-ICD system in parasternal midline in any CMR sequence used in the present study.

Due to the location of the S-ICD generator (Fig. 1), assessment of bSSFP-based cine-CMR images was predominantly impaired in the LV anterior, lateral and inferior wall segments (Fig. 2a-c). However, a switch to spoiled GRE-based cine-CMR substantially improved image quality in the majority of cases (Fig. 2d-f) and allowed an accurate assessment of cine-images in N = 17 (74%) patients with only limited artefacts. There was only one study in which assessment of cine-images was completely impossible due to very extensive device generator artefacts. The overall distribution of artefacts in cine-images is further illustrated in Fig. 3a.

Similar to cine-images, hyperintensity artefacts in conventional LGE-images were predominantly observed in the LV anterior, lateral and inferior wall segments caused by the S-ICD generator (Fig. 3b and Fig. 4a-c). Using optimized methods, LGE images with only limited artefacts were obtained in 13 (56%) patients. Noteworthy, the septal wall was assessable in all but one case, but poor image quality with substantial artefacts were documented in 10 (44%). Image optimisation by the wide-band technique was helpful in 15 (65%) of cases (Fig. 4d-f) and/or changing the phase encoding direction improved image quality in 9 (39%) patients.

Regarding T2w edema imaging, good image quality without any or only limited artefacts was observed in 16 (70%) patients (Fig. 5a-b). Aortic flow measurements using VENC-CMR was performed in 21 (91%) patients and was assessable without any relevant artefact issue in all of them (Fig. 5c-d). First-pass MPI at rest was performed in  16 (70%) patients and only half of these patients demonstrated perfusion images with acceptable image quality allowing valid visual assessment (Fig. 5e-g). Finally, additional 3D-CMR angiography of the thoracic aorta was performed in 15 (65%) patients and artefact-free reconstruction and assessment of the thoracic aorta was possible in all of them (Fig. 5h).

The major CMR findings in the present study are illustrated in Table 5. The average LVEF was 58 ± 10% whereas mean right ventricular (RV) ejection fraction (RVEF) was 58 ± 9%. An ischemic pattern of LGE was observed in 1 (4%) patient whereas a non-ischemic pattern of myocardial fibrosis was documented in 16 (70%); in 4 (22%) patients, the LGE pattern (in addition to T2w edema imaging) was suggestive of an inflammatory cardiomyopathy (including myocarditis and cardiac sarcoidosis).

A benefit in clinical decision-making was defined as illustrated in Table 6 and comprised (amongst others) a) identifying a new or confirming/changing a suspected diagnosis, b) identifying and locating myocardial scarring and assessing transmurality (helpful for optimal planning of ablation procedures and for risk stratification, e.g. assessment of the potential for future full functional recovery), c) identifying and locating myocardial inflammation for appropriate diagnosis of inflammatory diseases and subsequent targeted endomyocardial biopsy as well as therapy and d) targeted genetic analysis in those cases in which CMR was suggestive of a genetic cardiomyopathy (one patient with suspected cardiac laminopathy and another one with arrhythmogenic RV cardiomyopathy (ARVC) in the present study). Taken together, a benefit in clinical-decision making was documented in 17 (74%) patients in the present study.

One limitation of our study is the relative small number of patients. However, to the best of our knowledge the present study comprises the world’s largest S-ICD population undergoing CMR so far. The major goal of this study was to illustrate the feasibility, safety and clinical value of CMR in patients with a S-ICD. Future work needs to consider larger study groups and further upcoming imaging techniques such as T1- and/or T2-mapping. Finally, some previous studies suggested a correlation between minimal generator-heart distance and artefact size on LGE-images for patients with transvenous devices [13, 16]. Such a conceivable association could not be confirmed in the present study – probably due to the small study group.