Response:

In their letter, Drs Finitsis, Falcone, and Green summarize their retrospective review of medical records of 19 patients with “retained metallic fragments in the regions of the spine” who underwent MR examinations without adverse outcomes. MR studies provided diagnostic information in almost all cases that led to surgical intervention in three.

Although the objective of this work is quite laudable, the methods performed by Finitis, Falcone, and Green unfortunately do not permit one to reach the conclusions and recommendations made. A purely retrospective study was performed with no controls whatsoever. No explanation is provided of how the correspondents documented the lack of injury from MR exposure. Was any follow-up performed of these patients? Was there any investigation for possible subclinical internal injury to tissues adjacent to the metallic foreign bodies? Because the majority of patients seem to have had some prior injury, often in the area of the metallic fragment, it is quite possible that additional injury to these previously damaged tissues might go unnoticed, especially if no formal follow-up examination was performed. Further, is the correspondents' safety recommendation of metallic bullet fragments applicable to all field strengths? Is 2 T acceptable? How about a 4.7-T research system? Where do they draw the line—and why specifically there? It has been well documented (1–4) that there are innumerable types of “bullet fragments” from various sources—some (<20%) powerfully ferromagnetic (2), others weakly ferromagnetic, and others nonferromagnetic. On what basis are these data apparently ignored? How was the degree of ferromagnetism of the metallic bullet fragments in these 19 patients assessed to see if their conclusions would be applicable to other potential projectiles?

Perhaps one of the most important take-home lessons of MR safety is that a safe MR examination does not mean repeat MR will be safe for the patient. There are dozens of variables that can affect the outcome and safety of exposing a patient with metallic foreign bodies to MR. The degree of ferromagnetism of the metallic object impacts significantly on the safety of MR exposure. The dimensions and mass must be examined; whether an object is massive or linear, has a long-axis or spherical shape, raises concern for translational or rotational forces. The precise location of the object in or near the spine must be determined; is the object free within the thecal sac, anchored within the cortical bone, or embedded in the cord itself? The strength of the static magnetic field and static magnetic field gradient (spatial distribution of the static B_o magnetic field) should be identified. The field and field gradient traverse during scanning; the stronger the field and its associated gradient, the greater the translation (projectile) and rational (torque) forces, and presumably the risk to the patient. The rate the patient and metallic object move through the static field and field gradient increases risk. To state that “19 patients with bullet fragments in or near the spine were permitted to undergo an MR study, and no one got hurt,” and to conclude that it is safe to expose such patients to MR environments is fraught with peril and not scientifically sound. I would like to remind our readers of one patient who suffered an intraocular hemorrhage and unilateral blindness after inadvertent exposure to MR imaging at 0.35 T (5), and who was subsequently found to have had a 2- by 3-mm fragment of metal on his retina. What is not well known is that this adverse event occurred at the end of the study after three imaging sequences were successfully acquired (5). It is entirely possible that this patient might have exited the MR scanner without difficulty, just as he had entered it, and remained there for almost an hour without difficulty. Imagine the consequences if one would erroneously conclude that, because nothing untoward had occurred during the prior exposure, it would be safe to expose this patient to MR scanning once again in the future.

I am also personally aware of a patient who was placed in the bore of a high-field MR scanner. Site practitioners were not aware that the patient had a ferromagnetic Codman variangle aneurysm clip implant—the same type that was implicated in the death of another patient of intracranial hemorrhage during positioning in a high-field MR scanner (6). The former patient was removed from the scanner when the clip was identified on the initial scan of the brain, with fortunately no untoward outcome. Certainly one would assume that, because no injury had occurred in that case, it would be safe to prospectively place this patient again into another high-field MR scanner!

Clearly, deciding whether one should permit a patient into the bore or environment of an MR scanner should always be considered as a risk-benefit ratio to be assessed on a patient-to-patient basis. The potential risks, however, should be determined by carefully and prospectively performing studies based on scientifically sound methods. I would like to respectfully submit that the conclusions reached and broad recommendations suggested in this letter do not meet such criteria, and may inadvertently lead the uninitiated to draw inappropriately optimistic conclusions about the safety of a very unforgiving environment.

We, at the University of Pittsburgh Medical Center, weigh the potential benefits of an MR study for a particular patient in light of the potential risks this imaging technique poses. There is the possibility of substantial translation or rotation forces or motion of a metallic foreign body depending on where in the body it is located, and how and if it is anchored to cortical bone, as noted above. The decision to expose a patient to an MR environment should hinge on these variables.