Clinical utility of cardiovascular magnetic resonance imaging in patients with implantable cardioverter defibrillators presenting with electrical instability or worsening heart failure symptoms

The study was conducted in accordance with the local institutional review board and the standards of the University of Leipzig ethics committee. Written informed consent was obtained from all patients prior to CMR examination. All patients with an ICD referred for clinical management of electrical instability and/or worsening heart failure were seen by the cardiologist/electrophysiologist in charge and the decision to undergo CMR imaging was at the discretion of the clinician. Consecutive patients referred for CMR imaging with an ICD between June 2015 and December 2018 formed the study population. The CMR examination rigorously followed a previously published, standardized protocol [5]. All patients underwent chest x–ray prior to the CMR examination and in the presence of abandoned, epicardial or fractured leads, CMR imaging was not performed (i.e. patients were not considered for study inclusion). In addition, the time period between device implantation and CMR examination was > 6 weeks.

Directly prior to the CMR examination, an experienced electrophysiologist carried out a thorough ICD device interrogation in the scanner console room : battery status and sensing/pacing thresholds of all leads were documented and the device memory was evaluated for events (e.g. appropriately or inappropriately classified arrhythmias; date, number and kind of delivered antitachycardia pacing therapies etc.). Subsequently, devices were programmed for the CMR scanner environment: all tachyarrhythmia functions were switched off; MR–conditional labeled devices were programmed into the manufacturer–provided MR–safe mode settings; non–MR–conditional devices were programmed to pacing off, sensing–only (ODO or OVO) or to asynchronous pacing (VOO) depending on the intrinsic rhythm of the patient. The electrophysiologist and the programming device were present in the scanner console room throughout the entire CMR examination. Directly after completion of the CMR examination, patients underwent a repeat ICD interrogation in the console room and all devices were restored to their original settings.

A cardiologist with a high level of expertise in CMR imaging (I.P. or C.J., both > 20 years of experience and Society for Cardiovascular Magnetic Resonance and European Society of Cardiology level 3 trained) was present throughout the entire procedure for online evaluation of all CMR images and to guide or adapt a previously established CMR workflow/imaging strategy for device patients if needed to fully address the clinical indication [5]. Briefly, CMR imaging protocols were tailored to the clinical indication and diagnostic imaging modules were combined accordingly (i.e. ventricular function assessment, myocardial tissue characterization using wideband LGE imaging plus/minus T1- or T2-weighted blackblood turbo spin echo sequences, myocardial perfusion assessment using dynamic first-pass perfusion spoiled-gradient echo imaging during adenosine stress and conventional contrast-enhanced three-dimensional angiography scans). Cine imaging was performed using spoiled gradient echo (SGE) sequences with short–axis geometries being acquired before and long–axis geometries after contrast agent application. For LGE imaging, the wideband technique was employed (bandwidth of the inversion prepulse, 3.000 Hz; frequency offset, + 1.000 Hz) [6]. Both, cine and LGE imaging sequences were acquired during end-expiratory breath-holding. All CMR examinations were carried out on a 1.5 T CMR scanner (Ingenia, Philips Healthcare, Best, The Netherlands) using a 28–element array coil with full in–coil signal digitalization and optical transmission. Following current recommendations, whole–body specific absorption rate (SAR) was restricted to 2 watts per kilogram bodyweight. Patients were continuously monitored throughout the procedure based on vector–surface electrocardiogram (ECG), peripheral pulse oximetry, respiratory motion pattern, non–invasive blood pressure measurements as well as continuous visual and voice contact. If online evaluation of initial short–axis cine images demonstrated ≥12 non–evaluable left–ventricular (LV) segments due to severe generator–related artifacts, the CMR examination was considered non–diagnostic and terminated.

Quality of CMR images (i.e. cine, LGE, T1- and T2-weighted blackblood turbo-spin echo, dynamic perfusion and contrast-enhanced three-dimensional angiographic image sequences) was evaluated offline > 6 weeks later by two experienced CMR imaging experts (I.P., C.J.) in a consensus read; device–related image artifacts were assessed per myocardial segment of the LV and right ventricle (RV) following previously defined classification systems [9, 10]. Myocardial segments were considered evaluable if artifact–free in at least one standard cardiac geometry.

In all patients, clinical data were recorded at the time of the CMR examination and the indication for CMR referral was documented: electrical instability was defined as electrical storm (i.e. ≥ 3 separate arrhythmia episodes leading to adequate ICD therapies consisting of anti-tachycardia pacing (ATP) or shocks occurring within a 24-h period), ≤ 2 adequate ATP/shock(s), or sustained ventricular tachycardia (VT) below detection; progressive heart failure was defined as any increase in New York Heart Association stage of at least one functional class during the last three months. The CMR imaging results were determined by the consensus of two experienced CMR imaging experts (I.P., C.J.) within < 1 h following the CMR examination and the CMR–based diagnosis was documented. CMR–based change in diagnosis was defined as a new, previously not considered diagnosis (i.e. by clinical history, other cardiac imaging modalities etc.). CMR–based change in treatment decision constituted the sole responsibility of the electrophysiologist/cardiologist in charge and was defined as change in management consisting of any deviation from the referral to invasive procedures (e.g. electrophysiology study/ablation procedure, cardiac surgery, coronary revascularization etc.) or a change in medical therapy.

All analyses were done using SPSS (version 21, International Business Machines, Inc., Armonk, New York, USA). Continuous variables were given as mean ± standard deviation for normally distributed data; frequencies and percentages were used to describe categorical data. Differences between continuous and categorical variables were assessed using Student’s t–test and Chi–square test as appropriate. All tests were two–tailed and a p–value of < 0.05 was considered significant.

CMR examinations were carried out in 212 consecutive ICD patients; for patients who underwent repeat/follow-up examinations, only the first CMR dataset was considered for analysis. Four CMR examinations (4/212, 2%) were excluded due to non-diagnostic CMR image quality. Hence, the final dataset for analysis comprised of 208 CMR examinations of ICD patients. Data analysis was conducted for all patients and for the subgroups of primary versus secondary prevention ICD patients (n = 115 versus n = 93, respectively). Overall, 94 patients were equipped with an MR-conditional device (94/208, 45%; see Table 1); patients with MR-conditional devices were younger (58 ± 12 versus 61 ± 12 years; p = 0.023) and had a shorter mean time since ICD implantation (48 ± 43 versus 68 ± 54 months; p = 0.003). Referral indication consisted of documented VT (141/208, 68%), inadequate device therapy (15/208, 7%) or progressive heart failure symptoms; detailed patient characteristics are given in Table 1. In all ICD patients, device interrogation before and after the CMR examination yielded no significant changes (mean change of battery status 0.0 ± 0.1 V; lead threshold, 0.0 ± 0.3 V; lead impedance, − 7.5 ± 52.7 Ω; for all p = ns).

The following CMR diagnostic imaging modules were employed: cine imaging for ventricular function assessment (n = 208), myocardial tissue characterization (LGE, n = 206 + / - T1– or T2–weighted blackblood spin echo imaging, n = 49 or n = 55, respectively), resting or adenosine stress perfusion (n = 27) and contrast–enhanced 3D–angiography (pulmonary veins, n = 25 or thoracic aorta, n = 3). Results of CMR data are given in Table 2.

Overall, post–contrast cine imaging allowed for evaluation of 96.1 ± 6.1% of LV and 95.6 ± 6.6% of RV segments. For LGE imaging, LV and RV segmental evaluability amounted to 94.2 ± 7.8% and 95.8 ± 6.9%, respectively; T1- and T2 imaging resulted in 94.9 ± 7.2% and 91.3 ± 8.5% evaluable LV segments, respectively; perfusion imaging allowed for evaluation of 98.8 ± 2.2% of LV myocardial segments. In general, artifact occurrence was most common in the anterior LV segments; representative CMR imaging examples are provided in Figs. 1, 2, 3 and 4 and additional files 1–3 (video format: .mp4), respectively. Image quality of contrast-enhanced three-dimensional angiography of the pulmonary veins or thoracic aorta was unimpaired in all cases (100% evaluability).

Overall, CMR imaging data resulted in a change in diagnosis in 40% (83/208) of ICD patients with a significant difference between primary versus secondary prevention ICD patients (37/115, 32% versus 46/93, 49%; p = 0.01; see Table 3 and Fig. 5); based on CMR imaging data, the percentages reclassified mainly affected the following diagnostic categories: hypertensive heart disease (22/83, 27%), post-acute myocarditis (13/83, 16%), dilated cardiomyopathy (13/83, 16%) and normal (10/83, 12%). In addition to mere reclassification of underlying cardiac disease, newly discovered and important diagnostic findings consisted of detection of LV thrombus (9/208, 4%) and identification of LGE-defined fibrosis in non-ischemic cardiomyopathy (75/208, 36%) representing ventricular substrate as potential target for ablation therapy. Finally, for the group comparison of “electrical instability” versus “worsening heart failure” no significant difference for “change in diagnosis” was found (59/156, 38% versus 24/52, 46%, respectively; p = 0.288).

The information gained from CMR imaging resulted in an overall change in treatment in 21% (43/208) with a similar distribution in primary versus secondary prevention ICD patients (25/115, 22% versus 18/93, 19%, p = 0.67; see Table 4 and Fig. 5). The main effect of the CMR results on treatment change can be gauged when focusing on patients referred for an invasive treatment: after CMR, a total of 123 of 141 patients initially scheduled for a VT ablation procedure subsequently underwent VT ablation while in the remaining 18 patients (13%) the treatment categories changed to medical therapy (n = 14) or coronary revascularization (n = 4). In more detail, the CMR findings giving rise to deferral of VT ablation were evidence of myocardial ischemia (n = 4), acute myocardial inflammation/cardiac sarcoidosis (n = 4), newly detected LV thrombus formation (n = 5), absence of CMR-identifiable structural heart disease and, thus, subsequent opportunity to initiate class IC antiarrhythmic drug therapy (n = 2), extreme LV wall thinning after transmural ischemic myocardial infarction (< 2 mm, n = 1), newly diagnosed arrhythmogenic RV dysplasia (end-stage disease with extensive RV fibrosis formation, n = 1) and in one patient the presence of CMR evidence of substantial pericardial fibrotic tissue formation/epicardial substrate after coronary artery bypass grafting who was deemed not suitable for an epicardial approach. Likewise, in patients initially considered for coronary revascularization (n = 14), the results of CMR imaging led to the following treatment strategies: coronary revascularization (n = 5), medical therapy (n = 8) and pulmonary venous isolation procedure (n = 1, negative adenosine stress CMR ruled out progression of known CAD and atrial fibrillation was deemed to represent the cause of progressive dyspnea).

In addition, while n = 2 patients initially scheduled for device upgrade to CRT were not considered for this treatment after CMR detected extensive inferolateral infarction/scar tissue (thereby rendering CRT ineffective), another n = 3 patients were newly stratified for device upgrade based on accurate, CMR-derived volumetric determination of low systolic LV ejection fraction which had been rather overestimated by prior transthoracic echocardiography [12]. For the group comparison of “electrical instability” versus “worsening heart failure”, a statistically significant difference of “change in treatment” was noted (25/156, 16% versus 18/52, 35%, respectively; p = 0.004).

While the study cohort closely reflected the clinical spectrum typically encountered at a highly specialized, tertiary care center, the data as such may only be applicable to a similar clinical scenario. Some referral bias may be present in the study population, since the decision to request CMR imaging has entirely been left to the referring electrophysiologist/cardiologist and critically-ill patients were generally not considered for a CMR study.

Importantly, while at our institution CMR imaging in device carriers is routinely carried out with a high degree of CMR imaging expertise (> 20 years of experience) following a previously established workflow taking advantage of more recent CMR sequence developments and with adherence to current guideline recommendations, local expertise and/or scanner availability may vary considerably and must be taken into account when implementing CMR imaging of ICD patients as an integral component of the diagnostic workup [5, 6, 21].

On a per patient level, a 2% rate of “non-diagnostic” studies was determined (i.e. studies for which there was no interpretable CMR imaging data) and this may be considered the total rate of non-usable study information based on the proposed CMR imaging protocol in ICD patients; in case CMR imaging was carried out successfully, the rates of non-diagnostic segmental evaluability of regional wall motion and wideband LGE were 4 and 6%, respectively.