The safety of MRI scanning of pacemakers and ICDs: what are the critical elements of safe scanning? Ask me again at 10 000.

This editorial refers to ‘Magnetic resonance imaging of pacemakers and implantable cardioverter-defibrillators without specific absorption rate restrictions’ by M. Mollerus et al ., on page 947 The risks of scanning device patients that have devices not labelled ‘magnetic resonance imaging (MRI) Conditional’ have been well described, as have the proposed strategies to mitigate these risks. 1 Redundant monitoring, savvy pre-MRI reprogramming, and specific absorption rate (SAR) management are the cornerstones of the risk mitigation strategies when scanning device patients. 2 Two relatively recent position statements have been promulgated in Europe and North America guiding the clinician about the risk and conduct of MRI in device patients. 3 , 4 And while risks are generally acknowledged, the magnitude of the risk of MRI to the patient and the functionality of the device and leads are under active debate. The outcome of this debate is of no small consequence as it will significantly impact practice patterns (implant MRI conditional devices or not; extract retained leads or not), drive evidence-based guideline statements (who and how to scan), influence the conduct of clinical trials in the future (sizable costly human clinical trials, and or bench/animal testing or computer modelling for compatibility certification), and decide to what service burdens we commit to precious limited manpower (what level of supervision during MRI is required). 5 Over the past decade, a number of groups have reported a large number of safe MRI scanning events in a variety of device patients. The vast majority of these scans were conducted on pacemaker patients who were not pacemaker dependent. In fact, so many safe scans are reported that some investigators publically state ‘all of the concerns, for the most part, are almost hysterical concerns, and none of them are based on facts… Every kind of potential disaster that we have tried to research has all just ended up being an urban legend across the community or the literature’. 6 So, in short, for non-pacemaker-dependent patients, if the pulse generator is programmed ‘off’ prior to MRI, is the event rate of adverse effects during MRI so low that we can scrap much of what is currently in the guidelines as overkill? Given such bold statements 6 about the low risk of scanning non-pacemaker dependent device patients, one might even question whether it is necessary for monitoring and/or direct physician supervision during MRI as is mandated currently. Are all (or many) of the precautions we take during scanning device patients simply misplaced overblown concerns and a waste of time that add no incremental safety benefit to the patient's outcome because ‘MRI studies are safe in device patients’. 7 Just what, if any, are the critical safety practices that lead to a safe scan in non-pacemaker-dependent patients? In this issue of the Journal , Mollerus et al . 8 require that we question whether the SAR should be limited during MRI of device patients as has been recommended in the European position paper 4 and by others. 1 Mollerus et al . report the outcome of scanning pacemakers and implantable cardioverter-defibrillators (ICD) patients without limitations on peak SAR. After finding no clinically meaningful change in pacing thresholds, sensing, or impedances, the authors conclude that ‘MRI scans may be performed safely in selected patients up to a peak SAR of 3.2 W/kg… and that peak SAR level poorly predicts the safety profile of patients with pacemakers or ICDs who are exposed to MRI’. While the authors do not explicitly ask that the recommendations regarding SAR restrictions be reconsidered, principles of logic and parsimony require such reconsideration if what they say is true. Simply put, why should there be any SAR restrictions (beyond normal regulatory guidelines) if it does not incrementally add to the safe conduct of the scanning of device patients? A word or three about SAR, a term likely unfamiliar to many device implanters and clinicians who manage devices. Matter (human tissue, leads, pulse generators) can be warmed by the application of radiofrequency (RF) energy. The MR process applies trains of RF pulses at calibrated strengths and durations that alter the magnetic field within the MR bore. Parameters such as bore temperature, ambient temperature, relative humidity, air flow rate, perspiration, region and volume of the area irradiated and blood flow influence temperature rise in the patient. ‘A key variable in determining patient heating potential in an MR scanner is the power absorbed per unit mass of tissue, which is the SAR’. 9 Application of RF energy during MR scanning is adjusted to avoid producing a ‘core temperature rise in excess of 1°C and localized heating greater than 38°C in the head, 39°C in the trunk, and 40°C in the extremities’. 9 RF application (and thus SAR and heating) is decreased by changing MR parameters such as ‘increasing the RF repetition time, adjusting flip angles, and changing matrix size’. 8 Mollerus et al . noted that while decreasing SAR might be desirable, in ‘real-world situations… attempts to limit SAR levels may lead to suboptimal scans’. Provocatively, after reading the manuscript, one might conclude since ‘peak SAR level poorly predicts the safety profile of patients with pacemaker patients with pacemakers or ICDs’ (Mollerus) and limiting SAR may compromise image quality, we should abandon such attempts to limit SAR. In other words, why expose the device patient to the risk of MRI if you are not going to get a diagnostic scan? Importantly, other publications describing MRI of device patients (including references cited by Mollerus et al .) where SAR was limited to <2 W/kg (for safety reasons to reduce heating) failed to mention any compromise of image quality in the MR scans obtained. In fact, in multiple studies, ‘diagnostic quality’ studies 1 are the rule and not the exception with investigators reporting that despite limiting SAR below 2 W/kg ‘all clinically relevant MRI sequences necessary for diagnosis were performed’. 10 Additionally, the study by Mollerus et al . did not have a ‘low SAR’ control group to compare the unrestricted SAR images obtained in the study. How would one know if limiting all the scans acquired below the 2 W/kg level would not have resulted in adequate diagnostic image quality. No evidence is presented by Mollerus et al . that had they imposed such SAR limits, the scan quality would have been reduced. The use of the term ‘peak SAR’ by Mollerus et al . is also potentially confusing. Readers should know the term ‘whole-body SAR’ is different from ‘local SAR’ and ‘peak SAR’ and that these values are derived from complex calculations provided by manufacturers of scanners and ‘do not take into account any metallic conductors within the body, and therefore this estimation will fail to indicate a serious condition’. 11 While beyond the scope of this commentary, suffice it to say that ‘local hot spots’ (particularly around lead tips) may develop during MR imaging as the result of RF application despite a reassuringly low SAR displayed on the MR console. A sun may warm you slightly overall, but focused with a lens, the same sun may cause burns locally. The use of the term ‘peak SAR’ may give the reader of the Mollerus et al . manuscript the impression that a ‘worst-case scenario’ was evaluated. Nothing could be farther from the truth as a ‘worst-case scenario’ evaluation would involve, among other things, evaluating multiple other lead paths and positions as well as patients with other body compositions and size characteristics. It is impossible, with the multiple variables that lead to heating in patients with implantable devices, a human study will ever sufficiently evaluate the risk of heating a given pacing or ICD system. Regulatory authorities such as the U.S. Food and Drug Administration recognize this when they state ‘clinical or animal testing in an MRI scanner cannot characterize worst-case clinical conditions for lead heating’. 12 Going forward, computer modelling will be used to simulate worst-case scenario heating effects, with animal and human investigations being used to confirm preclinical results. The necessarily limited range of patient and device/lead characteristics of the author's (or any human study) work provides us with necessarily limited comfort in accepting their conclusions because failing to identify an adverse event is not equivalent to demonstrating safety—especially when only a limited number of patients are studied. The fact that ‘SAR as a poor indicator of MRI implant heating’ across different MR systems with different software is well documented and accepted. 11 , 13 Thus, in the conclusion of the Mollerus manuscript we find the standard disclaimer (found in all papers of this sort) that ‘the current study cannot be extrapolated to … other models of 1.5 T MRI scanners’. 8 However, investigations into the relationship of heating and SAR ‘support a linear relationship between SAR and heating within a given type of MRI system’. 13 Practically speaking, for your MRI system running a specific version of software on any given day, your patient (and the device/lead) will heat up more as you image the patient with MR sequences resulting in greater SAR. Incremental safety measures, if easily implemented (low or no cost, low or no service burden) should be implemented, when they result in safer outcomes even when the risk of adverse events may be low. It takes little time and effort to look both ways before you cross the street, but it's the safer thing to do even when the traffic is light. If a diagnostic quality MR image on a device patient can be obtained without difficulty at a lower SAR (<2 W/kg), why not lower the risk of heating complications, however small? I would recommend obtaining the image at the lower (safer) SAR value and seeing if a diagnostic image is acquired. If not, one can always repeat the scan with appropriate adjustments. Why salt your food before tasting it? The ‘keep the SAR low’ recommendations when imaging device patients should continue to be an important risk mitigation strategy and the Mollerus et al . manuscript does nothing to suggest a revision is required. Recently, a number of the measures suggested in the guidelines and position statements have come under scrutiny. For instance, the need to delay MRI of newly implanted devices has been questioned as pacing systems (pulse generators and leads) have little or no ferromagnetic material that would subject them to significant movement or lead dislodgement. 14 , 15 The wisdom and utility of performing defibrillation threshold testing on all ICDs after MRI has also been questioned. 15 Likewise, the utility of the post-MRI follow-up device checks might also be questioned. 4 If the only adverse effects of MRI on device patients occur during scanning and in the immediate post-scan interrogation none are detected (‘an entirely safe scanning event’), what is the utility of having the patient return for a post-MRI check, beyond the usual scheduled follow-up? At the very least, remote follow-up technology could be leveraged to diminish the service burden of a low or no yield post-MRI device check particularly in non-pacemaker dependent pacemaker patients who have devices that show no abnormalities at the immediate post-MRI check. Bringing the person back at 6 weeks to confirm the obvious seems unnecessarily burdensome, particularly since no one has put forth a plausible mechanism for how exposure to MRI might show deleterious ‘late effects’ in a device found to be entirely normally functioning post-MRI. Obviously, should problems be found post-MR, corrective action and or close follow-up is mandatory. How much monitoring during the scan is necessary? Because of interference during the scan, the EKG as a sole measure remains an insufficiently reliable means of monitoring the safe conduct of the scan. Should the EKG become uninterpretable during MRI, as it often does, I remain baffled as to how the physician will explain to the family of a patient who experiences a life-threatening arrhythmia why it was not interrupted earlier had something so simple as a pulse oximeter been added as a means of following the patient's progress during MR scanning. Redundant monitoring cannot be left to ‘physician discretion’! 16 Inadequate monitoring is precisely why patients with devices undergoing MRI have died. 17 It might be useful to recall that perhaps a mere 1500 or so scans have been reported on device patients in the medical literature. Surely, not enough safe scans have been done to declare all our previous concerns ‘hysterical’. 6 Numerous investigators scanning device patients have reported threshold and sensing changes, battery depletion, unexpected asystole, power-on-reset (POR) phenomenon, etc. even when appropriate risk mitigation strategies were applied. Undoubtedly, as more device patients are scanned, we will understand more fully the risks of scanning device patients. Accordingly, the safety recommendations and risk mitigation strategies for scanning non- MR conditional devices will be revisited and revised. What role relatively large MR scanning registries documenting the outcome of a hodgepodge of pacemakers and ICDs undergoing a variety of MR scans in different MR scanners will have in the formulation of future recommendations (or regulatory rulings) is open to debate. As recently noted, ‘a series of small changes can collectively lead to entirely new functions and, potentially, risks’ regarding devices and leads, 18 This is why uncontrolled evaluations of any number of devices (in a constantly evolving device and MRI market) will not answer the crucial question that every physician who wants to scan a device patient on any given day needs to answer: ‘What is the risk of performing this particular scan on this particular MR machine running this software on this patient with this particular device?’. The critical elements of safe scanning of device patients remain redundant monitoring, savvy pre-MRI reprogramming, and SAR reduction strategies. These are low-cost, trivially intrusive strategies that make scanning even low-risk non-pacemaker-dependent pacemaker patients safer. The consequences of POR during MR scanning to not only pacemaker patients, but also to particularly ICD patients, remain quite concerning even in the absence of pacemaker dependency and even when appropriate risk mitigation strategies are applied. 19 Those who question the need for MR conditional devices should bear in mind we have likely gone just about as far as we can in terms of making things safer with the monitoring, reprogramming, and SAR reduction strategies previously suggested of non-MR conditional devices. Skeptics are free to choose devices for their patients as they wish based on the myths and magical thinking that modern devices are MRI safe. Perhaps the sheer numbers of safe scans presented in the future will render my concerns for patient safety ‘hysterical’. Perhaps, I will not look both ways before crossing the street and begin salting my food before I taste it. Ask me again at 10 000 or 100 000. Conflict of interest: J.R.G. has received compensation from Sorin, Biotronik, and Boston Scientific, and Medtronic for speaking. J.R.G. has received compensation for clinical research from Medtronic. J.R.G. is a consultant to Medtronic, Biotronik and St. Jude Medical. J.R.G. owns stock equity in Medtronic and Boston Scientific.