Author + information
- Received April 13, 2016
- Revision received June 15, 2016
- Accepted June 17, 2016
- Published online October 4, 2016.
- aDivision of Cardiology, Department of Medicine, Johns Hopkins University School of Medicine, Baltimore, Maryland
- bDepartments of Radiology and Biomedical Engineering, Johns Hopkins University School of Medicine, Baltimore, Maryland
- ↵∗Reprint requests and correspondence:
Dr. Henry R. Halperin, Johns Hopkins Hospital, 571 Halsted Cardiology Division, 600 North Wolfe Street, Baltimore, Maryland 21287.
There is a growing population of patients with implanted electronic cardiac devices and a concomitant increase in the use of magnetic resonance (MR). There are theoretical safety risks posed to such devices by MR. However, there are now considerable laboratory data and clinical experience demonstrating safety in this setting, assuming appropriate device selection and patient monitoring. Herein, we review these data and our safety protocol and the new generation of devices that have been prospectively designed and tested to be safe for MR scanning, assuming certain conditions are met (i.e., devices that are MR-conditional). We also argue that the available data do not support a complete transition to implantation of MR-conditional devices.
The use of magnetic resonance (MR) has increased as an imaging modality in the diagnosis of a wide variety of conditions (1). Its growth can be credited to its high spatial resolution and soft tissue contrast in the absence of ionizing radiation or exposure to iodinated contrast agents. There is also a large and growing number of patients with implantable electronic cardiac devices, including pacemakers and implantable cardioverter-defibrillators (ICDs), with millions of people having received them in the United States (2).
With the growing use of MR, and a growing population of patients with devices, there is increased need for understanding the safety concerns and appropriate protocols for performing imaging in this population. There is increasing evidence that MR can be performed safely in patients who do not have magnetic resonance (MR)-conditional devices, but MR is still underused in these patients (3). In this review, we discuss the safety concerns of MR in patients with devices, the clinical evidence of safety, and the appropriate protocols for determining feasibility and safety in this setting.
Early experiences with MR and devices, without regard for technical considerations, led to significant adverse events (4). Studies demonstrated the ability of MR to cause radiofrequency heating of pacemaker leads, increased pacing thresholds, electronic dysfunction, and induction of malignant arrhythmia (5–8).
Factors influencing safety and potential adverse events with MR include the strength of the magnetic field, radiofrequency power, gradient power, magnet and device position, patient size, and specific features and components of the implantable device. Potential adverse events can be divided into mechanical, heating, and other electromagnetic effects and are summarized in the Central Illustration and Table 1.
The force generated by the magnetic field can theoretically cause device movement and dislocation. Static and gradient magnetic fields can induce force or torque on ferromagnetic components of the device. Modern leads are generally designed with minimal ferromagnetic components and are unlikely to be subject to movement (9). Some quantity of ferromagnetic material remains necessary, specifically in the battery and transformers needed for charging an ICD capacitor. Despite this concern, device movement is very unusual, and MR within 24 h of device implantation has been performed safely at our institution. Overall, the mechanical force and torque applied to devices by common MR clinical practice is on the order of that of physiological gravity (10).
Radiofrequency pulses used in MR lead to electric fields and radiofrequency energy deposition in body tissues (11). Device leads can exhibit electromagnetic coupling with the induced radiofrequency fields, leading to the “antenna effect,” in which there is resonant coupling of an electric field with a wire (12). In addition to potential tissue damage through deposition of heat at the lead-tissue interface, such effects can change pacing thresholds, cause loss of capture, or induce arrhythmia (8). Fractured, epicardial, and abandoned leads can be especially susceptible to heating effects (13). However, selection of an appropriate imaging landmark can substantially mitigate heating risk (14).
Electromagnetic effects other than tissue heating lead to other potential interactions with risk for adverse outcomes. Both the radiofrequency and gradient fields in the scanner can induce electrical current in leads. This can cause either oversensing or undersensing, which, in turn, may prevent appropriate pacing or antitachycardia therapies. Of note, ICD shocks likely will not be delivered inside an MR scanner because the magnetic field of the MR will saturate the transformer used for charging the capacitor, thereby preventing the capacitor from charging adequately (15). An additional possibility is activation of the device reed switch by the presence of the magnetic field. Reed switch activation typically leads to asynchronous pacing for pacemakers and disabling of tachycardia detection in ICDs.
Ineffective management of ventricular arrhythmia that would otherwise be sensed and treated by the device, therefore, requires vigilance for recognition and treatment by attending staff. Additionally, the electrical currents induced may potentially be strong enough to stimulate the heart and induce a ventricular arrhythmia (16). Electromagnetic interference during imaging can also lead to inappropriate tracking, pacing inhibition, unnecessary tachycardia therapies, and changes in device programming (17,18). In addition, MR interference can lead to a backup mode known as “power-on reset,” which can result in inhibition of pacing (19).
Prior to studies in humans, a number of in vitro and in vivo studies were performed to assess the mechanical, heating, and electromagnetic effects described previously. In pacemakers manufactured after 1996 and ICDs manufactured after 2000, in a 1.5-T magnetic resonance scanner, the maximal force was equivalent to 100 g, with a maximum torque of 90 g-cm (20). This quantity of force and torque is unlikely to move a device, especially considering surrounding tissues holding a pacemaker or ICD in place.
Clinical MR protocols were performed to assess temperature at pacemaker lead tips and ICD coils. Here, the temperature change was 1.0°C in vitro and <0.2°C in vivo (20). Temperature changes vary across scanners and MRI machines and with lead features, such as length, insulation, and proximity to scanner and coil (17). Nonetheless, these temperature changes are likely below those which would lead to clinically significant tissue effects, and the U.S. Food and Drug Administration (FDA) has accepted temperature rises <8°C as having no clinical significance.
In a canine model, MR imaging was performed 4 weeks after ICD implantation in 18 animals. Pacing thresholds and electrocardiogram amplitudes were unchanged, with the exception of a transient failure to capture in 1 animal. Subsequent pathology showed minimal tissue necrosis or fibrosis at the lead-tissue interface, which was consistent with control animals who did not undergo MR (20).
In a subsequent in vivo study in a canine model, the magnitude of induced current during MR was examined (21). The induced current was <0.5 mA, significantly below typical pacing thresholds. Capture thresholds with currents >30 mA were induced when 4 or more lead loops were added to increase the circuit area, although this is exceedingly unlikely to be encountered clinically. Other in vitro studies demonstrated that the risks of an induced current may be greater with abandoned leads and leads connected to a pulse generator with loss of hermetic seal at the connector (22).
Over the past 2 decades, a number of studies have investigated the safety and feasibility of MR in patients with pacemakers and ICDs. Of the theoretical harms that could be caused by MR, movement of the device by the force of the magnetic field and heating effects have largely been nonissues. Electromagnetic effects on pacemaker programming and function are summarized in the following text.
Clinical trials of MR in patients with permanent pacemakers are summarized in Table 2. Of note, these trials are all studies of patients with current devices that were not approved as MR-conditional and included patients who were pacemaker-dependent. In permanent pacemaker patients, initial studies investigated safety in 0.5-T scanners in 76 patients. The most commonly observed adverse events were reed switch activation leading to asynchronous pacing and decreased battery voltage following the study (23,24). Subsequent studies in 1.5-T scanners have included over 800 patients and demonstrated minor changes in pacemaker function, such as depleted battery and altered pacing thresholds but no significant adverse patient outcomes (6,7,19,25–33). Although statistically significant changes in pacing thresholds and lead impedances were observed in some studies, these changes were not at clinically significant levels. Decreased battery voltage has also been observed and highlights the need for monitoring of battery status before and after MR. Power-on-reset events were also observed, which carry a theoretical risk of changing a pacing mode from asynchronous to inhibited and therefore inappropriate inhibition of pacing. In smaller numbers of patients, including 1 review of 44 patients, studies have also been performed in 3.0-T scanners without significant adverse outcomes (34).
Likewise, a number of studies have been performed regarding MR in the setting of an ICD, as summarized in Table 3. Across several studies including over 300 patients in 1.5-T scanners, there have been no significant adverse clinical outcomes (19,25,27,30,32). The most common issue has been statistically significant but clinically insignificant changes in lead parameters. To assess the impact of serial examinations, 10 patients with ICDs were followed for 3 serial scans over a 1-year period, which demonstrated no changes in battery or lead parameters (35).
In a review of numerous studies of MR in patients with devices, no mortalities were reported (36). Additionally, in a retrospective study comparing device patients who received MR with those who did not, there were no differences between groups, suggesting that some minor changes in lead parameters may be due to random fluctuations in device function (37). At our institution, >2,500 scans have been performed on patients with implantable devices following the safety protocol described below, of which an initial series of 555 was previously published (19). The most significant events were power-on-reset events in <2% of patients and transient asynchronous pacing from reed switch activation. Lead parameters were largely stable, and there were no episodes of inappropriate activation or inhibition of pacing. In our experience, only 1 scan was stopped prematurely due to patient sensation of device movement, but this occurred in an older model ICD in which we currently do not recommend MR (19). Patients who have had power-on-reset events have been able to return for subsequent MRs. The totality of the data in patients with non-MR-conditional devices undergoing MR supports the fact that this can be done safely, without adverse patient outcomes and with minimal effects on device function.
Johns Hopkins University Safety Protocol for MR with Pacemakers and ICDs Not Labeled as MR-Conditional
The safety protocol at our institution has been used in over 2,500 scans in patients with ICDs and is outlined in Figure 1. First, a risk/benefit decision is required by the patient and clinician to weigh the necessity of MR versus potential risks. Informed consent is routinely obtained. In our experience, there have not been medicolegal issues related to MR of non-MR-conditional devices.
Due to concerns of electromagnetic interference with older generators, pacemakers implanted before 1998 and ICDs implanted before 2000 are excluded. Patients with temporary leads, epicardial leads, abandoned leads, and implants within 4 weeks are also excluded and are discussed in greater detail in a separate section.
For all scans involving non-MR-conditional pacemakers and ICDs, a cardiac electrophysiologist or cardiac nurse trained in advanced cardiac life support with implantable device programming experience is present to assist with any potential programming issues. At our institution, an electrophysiologist was present for the first 100 scans. Since then, a nurse has been present, with a physician on call in case of emergency. An electrophysiologist has been present when performing scans in patients in whom we have recommended against performing MR (such as those with a newly implanted device or abandoned leads), but the clinical scenario suggested that the benefits outweighed the risks.
Vital signs are recorded, as are device settings and lead parameters. The mode of pacing is then adjusted, depending on whether or not the patient is pacemaker-dependent. In order to mitigate risk of inhibition of pacing due to detection of radiofrequency or gradient signals, pacing is set to asynchronous pacing (VOO or DOO) in pacemaker-dependent patients. To avoid inappropriate sensing of radiofrequency pulses and therefore inappropriate tracking in patients who are not pacemaker-dependent, pacing is set to a nontracking mode (VVI or DDI). For similar reasons, settings such as rate response, premature ventricular contraction response, ventricular sense response, and atrial fibrillation response are deactivated. To minimize asynchronous pacing and risk of inducing arrhythmia, magnet mode and noise reversion are deactivated. For ICDs, tachyarrhythmia monitoring is deactivated to avoid battery drainage and tachyarrhythmia therapies are deactivated to avoid antitachycardia pacing or shocks.
During the MR, electrocardiograms are recorded and blood pressure and oxygen saturation are monitored. Because no specific imaging sequences are known to increase patient risk, any imaging protocol can be performed, as needed. As discussed later, additional sequences or adjustment may be made to limit image artifacts from the device. Post examination, the original programming is restored, and lead parameters are compared to baseline values and adjusted, if necessary. A follow-up device interrogation is then performed at 3 to 6 months.
Effect of Pacemakers and ICDs on Image Quality
In addition to potential safety risks, electrically conductive devices, such as pacemakers and ICDs, which may contain some ferromagnetic materials, have the ability to distort MR images. When the implantable device is not in the field of view, as in nonthoracic imaging, it has no impact on MR image quality. In thoracic imaging, the implantable device may distort the magnetic field, leading to image artifacts, such as distortion, bright or void signal areas, and decreased fat suppression. The area of artifact or distortion is significantly greater with ICD than with pacemaker generators (38). As artifacts are more likely to be present on inversion recovery sequences, correlation with other pulse sequences is critical to avoid overinterpretation of artifacts. Imaging features to reduce potential artifacts include use of image planes perpendicular to the generator, reduced echo time, and use of spin echo and fast spin echo sequences. Recent work has demonstrated new tools to reduce image artifacts in the setting of ICDs, such as with increased bandwidth of inversion and excitation pulses (39).
Special Clinical Situations
In our experience, it is safe to perform thoracic and nonthoracic MR in patients with implantable loop monitors. This has been reported previously, with no impact on device function, battery life, or patient symptoms (40). Of note, it is important to interrogate the device after scanning to clear episodes of arrhythmia that are attributable to artifacts from the MR. Commonly used implantable monitors, such as the Reveal (Medtronic, Minneapolis, Minnesota) and Confirm (St. Jude Medical, St. Paul, Minnesota), are labeled MR-conditional, reflecting their safety in this context.
Temporary transvenous pacemakers
Given the nonfixed and less stable nature of transvenous temporary pacemakers, they are more prone to movement. In an in vitro model, temporary leads heated to over 15°C and up to temperatures that could theoretically cause significant tissue damage (5). Such leads are also theoretically more susceptible to electrical artifacts than permanent pacemakers. At our institution, MR has been performed safely with an active fixation lead attached to an externalized permanent pacemaker.
Recently implanted devices
With recently implanted devices, there is risk of spontaneous lead dislodgement unrelated to MR. Theoretically, the generator may be dislodged due to the MR-related force and torque, but this risk is exceedingly low, even immediately post implantation. In 1 study, 8 patients underwent MR at 1.5-T within 6 weeks of device implantation. None of the 8 patients was pacemaker-dependent or receiving a generator change. There were no clinically significant complications (41). At our institution, we typically avoid MR within 4 weeks of implantation but have safely performed these scans within the 4-week time period in specific clinical emergencies. Given the relatively limited experience in this setting, we recommend waiting until 4 weeks after implantation but acknowledge that imaging is possible within this time frame if clinically necessary.
Epicardial and abandoned leads
Epicardial and abandoned leads are of greater concern with MR, given the increased risk of heating and tissue damage. Abandoned leads have been shown to heat to a greater degree than pacemaker-attached leads (13). Epicardial leads do not have the advantage of being cooled by blood flow. Given these concerns and the lack of clinical experience, we currently do not recommend imaging in the presence of either epicardial or abandoned leads, unless there is an extremely compelling clinical necessity.
Medical devices are currently labeled by the American Society for Testing and Materials as MR-conditional when an item demonstrates no known hazards in an MR environment with specified conditions. Of note, although older devices are not labeled MR-conditional, their safety under specified conditions has been demonstrated. Nonetheless, device manufacturers are currently developing MR-conditional systems that are at different stages of FDA approval.
Efforts to make devices more MR-compatible have included advances in both hardware and software. MR-conditional devices have reduced amounts of ferromagnetic material, further limiting the risk of movement or other mechanical issues. Filters have been introduced to limit transfer of certain frequencies and dissipate energy, resulting in reduced damage to the internal circuitry and less inappropriate sensing. Lead geometry has been redesigned, limiting conduction of specific radiofrequencies and thereby reducing tissue heating. Reed switches have been altered or replaced with Hall sensors, theoretically reducing the risk of inappropriate activation by the magnetic field. The new generation of MR-conditional devices also has a MR mode that allows for simplified enabling of asynchronous pacing when appropriate, deactivation of advanced pacing functions, and simple restoration of prior settings.
In a major trial of the first MR-conditional system (Revo EnRhythm MRI pulse generator and CapSureFix model 5086 MRI leads, Medtronic), there were no adverse patient outcomes, and the FDA subsequently approved the device as MR-conditional (42). Despite this, some safety concerns have been raised with this system, including issues with cephalic vein access and lead dislodgement (43,44).
A second-generation system (Advisa MRI pulse generator and CapSureFix 5086 leads, Medtronic) showed no significant adverse events in clinical testing and has also been approved by the FDA as MR-conditional for full-body scanning. In a study of 36 patients at a single center, using the EnRhythm/Advisa SureScan pacemakers, patients underwent MR (45). No MR-related adverse events were recorded, although clinically-insignificant changes in capture thresholds were noted. The previously not MR-conditional CapSureFix Novus 5076 (Medtronic) leads have since been studied and show a good safety profile when connected to the Advisa SureScan (Medtronic) MR-conditional generator, resulting in approval of the 5076 lead as MR-conditional in this setting (46).
The Entovis pacemaker system (Biotronik, Berlin, Germany) was evaluated in the ProMRI/ProMRI AFFIRM study, which showed no adverse events in thoracic and lumbar imaging (47,48). This resulted in FDA approval as MR-conditional in 2014. Also, MR-conditional ICD systems have been studied and are now FDA approved but are not discussed herein.
Moving forward, clinicians will need to decide whether to implant MR-conditional or older devices, sometimes referred to as “legacy” devices. Presently, there is a lack of consensus or guideline statements directing future device selection. Some clinicians have argued for the exclusive implantation of MR-conditional devices on the basis of safety, supervisory, administrative, and economic advantages (49). Herein we argue for continued placement of legacy devices and the need to widely implement protocols for MR imaging of legacy devices.
There is now a significant amount of clinical data demonstrating the safety of MR with legacy devices in appropriate settings. Many of the observed changes following MR have been of minimal clinical significance. Although the MR-conditional devices have undergone significant testing and also have robust safety data, the differences in safety are not significant enough to justify a uniform change in practice. It should also be noted that there is more experience with the long-term safety of legacy devices with regard to pacemaker and ICD risks other than MR, such as lead fracture or dislodgement.
A common argument for complete adoption of MR-conditional devices is the need for less intense patient supervision in the MR environment. This is short-sighted in 2 respects. First, although MR-conditional devices have specific MR modes and less risk of clinically significant or insignificant events, they still require reprogramming. Some form of skilled clinical personnel will be required whether a patient has an MR-conditional or a legacy device. Second, even if a wholesale change to MR-conditional devices were made, there would still be a large population of patients with legacy devices in need of MR at some point in the coming decades. To have a policy of imaging only MR-conditional devices will deny a critical diagnostic modality to a large number of patients. To truly improve patient access to MR, institutions should transition to imaging patients with all devices for which there is safety experience, rather than limiting MR to those with MR-conditional devices.
Admittedly, in MR using legacy devices, there are lingering administrative issues related to compensation, medicolegal consequences, and compliance. Although these are legitimate concerns, they do not justify denial of MR to those with legacy devices. Off-label use of drugs, devices, and diagnostic modalities is commonplace in modern medical practice. Just as with other therapies, a needed diagnostic test cannot be routinely refused because it is off-label when there is such robust experience and safety data.
Finally, there is the issue of cost. An argument could be made that the increased cost of MR-conditional devices is offset by less need for patient monitoring and less possibility of compensation or medicolegal issues. As mentioned previously, however, supervision costs will not significantly differ. Additionally, although there is abundant data to suggest safety with legacy devices and MR, there are no such data to support economic disadvantages, beyond theoretical concerns.
Although MR-conditional devices represent an important advance in the field, their advantages over legacy devices do not justify a complete transition in practice, considering differences in cost. Additionally, regardless of whether a center moves to MR-conditional devices, the population with legacy devices will continue to be large enough to justify providing them access to MR for years to come.
Conclusions and Future Directions
Issues of implanted devices and MR compatibility are becoming more frequent. Despite many theoretical safety risks, there is extensive data to support safe use, given appropriate precautions. Although a new generation of devices labeled MR-conditional is coming to market, to some extent, all devices implanted after 2000 are safe for MR, assuming appropriate conditions are met. It is critical that patients with these older devices are not excluded from MRs when clinically appropriate. In fact, some older leads have been retested and have met criteria for being relabeled as MR-conditional. All centers performing MR in patients with devices need a dedicated plan or checklist, with appropriate monitoring by skilled personnel.
Manufacturers will continue to devise implantable devices with less susceptibility to alterations in function in MR magnetic and gradient fields. The ultimate goal is to have devices that are MR safe, such that there would no longer be any need for additional precautions. There is also increased use of 3-T MR systems, and, although there is some clinical experience in this context, further study is required, as most of the safety data are in the setting of 1.5-T scanners.
Dr. Nazarian is a consultant for CardioSolv, Medtronic, and Biosense Webster; and is principal investigator for research grants to Johns Hopkins University from Biosense Webster and U.S. National Institutes of Health (R01 HL094610). Dr. Halperin has received royalties from Imricor. Dr. Miller has reported that he has no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- Food and Drug Administration
- implantable cardioverter-defibrillator
- magnetic resonance imaging
- Received April 13, 2016.
- Revision received June 15, 2016.
- Accepted June 17, 2016.
- American College of Cardiology Foundation
- Smith-Bindman R.,
- Miglioretti D.L.,
- Larson E.B.
- Levine G.N.,
- Gomes A.S.,
- Arai A.E.,
- et al.
- Martin E.T.,
- Coman J.A.,
- Shellock F.G.,
- et al.
- Sommer T.,
- Naehle C.P.,
- Yang A.,
- et al.
- Luechinger R.,
- Zeijlemaker V.A.,
- Pedersen E.M.,
- et al.
- Irnich W.
- Roguin A.,
- Schwitter J.,
- Vahlhaus C.,
- et al.
- Nazarian S.,
- Beinart R.,
- Halperin H.R.
- Ainslie M.,
- Miller C.,
- Brown B.,
- et al.
- Roguin A.,
- Zviman M.M.,
- Meininger G.R.,
- et al.
- Nazarian S.,
- Roguin A.,
- Zviman M.M.,
- et al.
- Mollerus M.,
- Albin G.,
- Lipinski M.,
- et al.
- Junttila M.J.,
- Fishman J.E.,
- Lopera G.A.,
- et al.
- Kamperdijk E.W.,
- van Nieuwkerk E.B.,
- Verdaasdonk M.A.,
- et al.
- Rashid S.,
- Rapacchi S.,
- Shivkumar K.,
- et al.
- Friedman H.L.,
- Acker N.,
- Dalzell C.,
- et al.
- Elmouchi D.A.,
- Rosema S.,
- Vanoosterhout S.M.,
- et al.
- Bailey W.M.,
- Mazur A.,
- McCotter C.,
- et al.,
- ProMRI Study Investigators
- Gimbel J.R.,
- Passman R.,
- Kanal E.
- Strach K.,
- Naehle C.P.,
- Mühlsteffen A.,
- et al.
- Technical Considerations
- Preclinical Testing
- Clinical Testing
- Johns Hopkins University Safety Protocol for MR with Pacemakers and ICDs Not Labeled as MR-Conditional
- Effect of Pacemakers and ICDs on Image Quality
- Special Clinical Situations
- MR-Conditional Devices
- Conclusions and Future Directions