This article presents a review of the current literature on the safety of MR Contrast Agents.
Val M. Runge, MD Rosenbaum Professor of Diagnostic
Radiology, University of Kentucky, Lexington
By almost any measure, injectable gadolinium-based magnetic
resonance (MR) imaging contrast agents are among the safest
exogenously administered compounds. Administered at volumes well
below those associated with iodinated contrast agents, and cleared
by the kidneys with no demonstrated nephrotoxic effects at approved
doses, these compounds are often considered innocuous. Several
published reviews attest to the excellent safety profile of the
gadolinium agents as a class.
These studies show the gadolinium chelates to be safe and well
tolerated. Minor adverse reactions do occur, and the articles point
out that radiologists and technologists should expect to see nausea
in a small percentage of patients and hives in a lower percentage
of patients, regardless of the agent used. The published reviews
also show that the agents cannot be differentiated on the basis of
adverse reaction rates.
Published case reports in the literature also attest to that
fact that idiosyncratic, anaphylactoid events can be associated
with the administration of gadolinium contrast agents.
In fact, at least one published report links the use of a
gadolinium contrast agent (Magnevist) with the death of a patient.
Overall, the literature on the safety of MR contrast agents is
sparse and occasionally misleading. This reflects radiologists'
satisfaction with the performance of this class of agents as well
as a general lack of safety studies sponsored by the contrast agent
manufacturers and distributors. In some instances, inappropriate
attempts have been made to differentiate various gadolinium agents
on the basis of survey data or individual published adverse event
reports. Therefore, a more comprehensive review of the available
data on the safety of MR contrast media is warranted.
There are currently four agents approved for use in the United
States (figure 1). The main chemical difference found in these
agents is in the molecular structure, or chelate, which can be
linear (Magnevist, Omniscan, OptiMARK) or macrocyclic (ProHance,
Dotarem). The chelate (from the Greek root "chelos" or claw) holds
the gadolinium ion tightly and securely and ensures rapid excretion
via the kidney. Since gadolinium is known to be highly toxic in a
noncomplexed state, the chelate is thus the basis for the
acceptable safety of this group of contrast agents. The stronger
bonds of the macrocyclic type agents show less tendency to release
elemental gadolinium over time.
While this structural difference appears to have little to no
impact on the acute safety effects of these agents, chelate
stability has potential safety implications in cases of delayed
clearance (renal impairment) or repeated injections (patients
receiving multiple follow-up examinations, such as patients with
treated tumors or multiple sclerosis). Fritz and colleagues
described the potential for dechelation of the gadolinium chelate
complex in vivo using a model involving liposomal gadolinium
However, the clinical consequences of free gadolinium require
The other key difference in the physicochemical properties of
these agents is related to the charge of the molecule (table 1).
Agents such as Magnevist or Dotarem (Gd-DOTA, available only in
Europe and South America, but not in the United States) with a net
negative charge must be formulated with positively charged ions to
assure a net zero charge in the final formulated product. Agents
such as ProHance and Omniscan with neutral molecules need no
additives to be electrically neutral. Thus, the osmolality
(particles per unit of solution) is much lower for the non-ionic
agents (e.g., 630 mOsmol/kg for ProHance vs. 1,960 mOsmol/kg for
Magnevist). Clinically, this difference translates into increased
viscosity of the higher osmolar agents, and a greater potential for
local irritation upon injection and tissue damage in cases of
Adverse Reaction Rates
Determining the "true" rate of adverse events associated with
gado-linium agents is challenging for several reasons. A certain
number of adverse reactions are likely related to anxiety produced
by the MR imaging examination itself. In their review of the
literature on non-contrast MR imaging, Melendez and McCrank
determined that up to 30% of patients experience "anxiety-related
reactions" when studied by MR imaging.
These reactions range from annoyance to claustrophobia or panic. In
addition, patients undergoing MR imaging exams are often severely
ill and exhibit a host of symptoms usually associated with their
disease processes. Thus, determining what reactions might be
attributable to the contrast agent used can be problematic. The
usual assumption is that a fraction of the adverse events recorded
in clinical studies are actually related to the agent
The most comparable safety data sets for the available
gadolinium chelates are found in the package inserts. The approved
labeling for each gadolinium agent contains a description of the
safety profile of the agent as evaluated under controlled
conditions during the clinical trials conducted for FDA
registration. A quick evaluation of the package inserts for the
four gadolinium chelates currently marketed in the United States
shows that the type and incidence of adverse reactions with these
agents are similar (table 2). All of the gadolinium agents approved
in the United States cause nausea at a rate of approximately 1% to
3%. The reported incidence of nausea with Magnevist is 2.7%;
ProHance, 1.4%; Omniscan, <3.0%; and OptiMARK 3.2%. It is
important to note that although the rate of adverse events is
slightly higher with OptiMARK, this most likely represents a trend
to increased adverse event reporting in clinical trials, as this
agent was the most recently tested and approved. Overall, the
evidence suggests that there is no clinically relevant difference
in adverse events between the agents in use in the United States
In addition to nausea, hives is also a relatively common adverse
reaction, and it is an event that we see more frequently the more
we look. Typically, hives will begin to develop 5 to 10 minutes
after injection. The presentation clinically is often that of four
or five large hives (1 cm diameter) scattered over the torso.
Generally, they resolve without treatment and disappear within 4 to
5 hours. Overall, only about 1 in 200 patients develops hives. Most
physicians and technologists have never seen a hive caused by a
gadolinium chelate. The reasons are multifactorial: symptoms are
minimal, patients are often released prior to maximal development,
and a physical exam is almost never done following contrast
The first gadolinium agent to be available clinically was
Magnevist, which was approved in the US in 1988. Well over five
million injections have been administered worldwide since the first
approval of this agent.
The safety profile of this agent in clinical practice is generally
considered to be excellent.
The first study using a gadolinium contrast agent in humans was
published by Laniado in 1984.
In this study, Magnevist (Gd-DTPA) was well tolerated in 20 healthy
volunteers. One of the earliest large-scale safety studies of a
gadolinium contrast agent reported in the literature was published
by Carollo and colleagues in 1990.
In this study of 225 consecutive patients receiving Magnevist, 17%
of patients experienced an adverse event, with 5% reporting events
that were considered to be related to contrast administration. The
most commonly reported adverse events were injection site pain
(5.7%), headache (5.3%), and nausea (1.7%). Also in 1990, Goldstein
and colleagues published the safety profile of Magnevist as
assessed in US clinical trials.
As expected, the most frequent adverse events in this series of
1,068 patients were headache (6.5%), injection site discomfort
(3.6%), and nausea (1.9%).
In one of the largest sponsored post-marketing safety studies,
Nelson et al reported on 15,496 patients given Magnevist.
In this prospective study 2.4% of patients reported an adverse
event; the most common being nausea and headache, each with a
reported incidence of about 0.4%. Interestingly, slightly less than
half the adverse events reported in this study were recorded at
least 1 hour after drug administration (but within the first 24
hours after injection). Since in clinical practice most patients
leave the MR suite soon after their exam is completed, this study
suggests that the overall rate of adverse events may be higher than
clinicians usually suspect.
A rigorous prospective study of the safety of one gadolinium
agent (ProHance) at a single center was published by Hieronim and
colleagues from the University of Pittsburgh in 1995.
In this study, the overall rate of adverse events after ProHance
administration in 3,558 patients was 2.1%, a rate similar to that
reported for Magnevist (2.4%) in the prospective study by Nelson et
Hieromin also reports earlier data from a retrospective chart
review to document the reported safety of Magnevist. The overall
rate of recorded adverse events with Magnevist in the retrospective
portion of the study was 1.3%. Since prospective data collection is
known to provide more comprehensive and accurate data than
retrospective data collection, the rates reported for ProHance
(prospective, 2.1%) and Magnevist (retrospective, 1.3%) should be
considered equivalent. For both agents the most frequently reported
adverse event was nausea with and without emesis, occurring in less
than 1% of patients dosed.
An important study (and one that is often quoted out of context)
was published by Murphy et al of the University of Michigan in
This study was a single-site, retrospective study of adverse
reactions to gadolinium contrast agents at the University of
Michigan. This study was neither designed nor presented as a
comparative study. The authors reported results from approximately
20,580 patients given Magnevist, approximately 372 patients given
Omniscan and less than 50 patients given ProHance. The researchers
found 36 recorded reactions (0.2%). A total of 15 patients had
nausea or vomiting, 12 had hives or diffuse erythema, 7 experienced
moderate respiratory symptoms, and 2 patients experienced cases of
life-threatening respiratory distress. Both of the severe reactions
were in patients who received Magnevist. Murphy reported 29
reactions after Magnevist (0.14%), 5 reactions after ProHance
(10.42%), and no reactions after Omniscan. The authors conclude
that although gadolinium is a safe contrast medium, anaphylactoid
reactions do occur.
This study probably underestimates the true incidence of adverse
events after Magnevist because data were collected retrospectively
and because injection site reactions and delayed reactions were not
included in the analysis. Nevertheless, due to the large number of
patients involved, the study has considerable value for assessing
routinely reported reactions to Magnevist (over 98% of all patients
in this analysis received Magnevist). For Omniscan, and especially
for ProHance, the number of patients involved is insufficient for
making any judgment about the performance of these agents in
routine clinical practice. For example, the reported reaction rate
for ProHance in Murphy's small sample is about 30% to 50% higher
than that found in well-controlled clinical trials with ProHance,
and five times higher than the rate reported by Heironim and
colleagues (2.1%) in their prospective trial with ProHance in over
A later study, published by Murphy in 1999, is also often quoted
out of context when discussing adverse reactions.
For this study, Murphy and colleagues conducted a survey of
American Society of Neuroradiology (ASNR) Fellowship Directors to
assess the reported prevalence of adverse reactions to
gadolinium-based MR contrast agents. A questionnaire was mailed to
105 directors of ASNR fellowship programs, with a response rate of
about 50%. Program directors were questioned on the number of
procedures performed annually utilizing various MR contrast media.
They were also asked to assess (from memory or a retrospective
review of records) adverse events to gadolinium contrast agents and
categorize them as allergic or non-allergic.
Respondents reported administering over 867,000 doses of
Magnevist, and recalled 314 (0.046%) non-allergic reactions, 107
(0.016%) mild allergic reactions, 28 (0.004%) moderate reactions,
and 5 (0.0007%) severe reactions. For Omniscan and ProHance the
number of administered doses was significantly less. About 74,000
doses of Omniscan were administered with 11 (0.015%) reported
non-allergic reactions, 12 (0.016%) mild reactions and no moderate
or severe reactions. For ProHance, respondents reported
administering about 64,000 doses, with 171 (0.27%) non-allergic
reactions, 49 (0.077%) mild allergic reactions, 29 (0.045%)
moderate reactions, and 11 (0.017%) severe reactions reported.
The most important take-home point from this survey is that, as
a class, the MR contrast agents are thought to be among the safest
intravenously administered drugs ever developed. The total reported
reaction rate was well under 1%, indicating that these drugs are
perceived to be extremely safe.
As a record of physician opinion this survey is very
informative. As a scientific study, however, this publication
suffers from severe limitations. First, it is a retrospective study
in which participants were asked to recall the number of doses
administered and the number of adverse reactions they experienced.
Therefore, the survey is subject to reporting bias and
under-reporting. In addition, there was no verification of the
responses with the source data. Therefore, neither the denominator
(administered doses) nor the numerator (reported reactions) can be
assumed to be accurate. Naturally, this makes calculating adverse
event rates highly suspect.
A second problem with the survey is that of unequal distribution
of patients in the different study groups. Over 90% of the
participants were injected with only one agent, Magnevist. In
addition, the data for another agent (ProHance) is heavily skewed
by a single center noted for "meticulous record keeping" which
administered 26% of the reported ProHance doses but accounted for
over 80% of all adverse events reported for ProHance (this center
also reported a 7.5% reaction rate with Magnevist, well above the
rate reported from other centers). This center reported 10 of the
11 severe reactions attributed to ProHance and 1 of the 5 severe
reactions attributed to Magnevist. It is clear that the inclusion
of this center biased the data provided for ProHance in a
unfavorable way. The authors acknowledge this problem in their
discussion of the results, but then compound it by attempting to
perform statistics on the survey data to calculate an odds ratio
for experiencing an adverse event to a given gadolinium agent.
Given the unreliability of the source data and the skewing effect
of the one center reporting high adverse event rates, attempting to
perform statistics on the survey data is probably not
As the authors state in the conclusion of the paper, due to the
numerous limitations of the design and analysis of the study, "the
results can therefore be interpreted as the opinions of only a
small number of experts and do not necessarily represent a sample
of practice patterns used throughout the world." Despite its
obvious shortcomings, this survey has been occasionally quoted to
suggest differing rates of adverse reactions with different agents.
However, due to the problems reported above, performance
comparisons between agents using this dataset are ill-advised. The
balance of the literature suggests all available gadolinium
chelates have a similar safety profile when used as directed.
In 1995, an overview of the safety of ProHance in the U.S.
clinical trial experience was published by Olukotun.
This study reported on over 1,700 patients and volunteers given
ProHance at doses from 0.05 to 0.3 mmol/kg. As expected, the
incidence of nausea was very low (1.4%). All other adverse
reactions other than taste perversion were recorded in less than 1%
of subjects dosed.
This study also reported no statistically significant difference in
adverse reaction rates between bolus and infusion of ProHance.
A larger study, published in Europe, compiled worldwide data and
included over 2,000 injections of ProHance.
Again the data fall within the expected range: nausea at about
1.5%, taste perversion slightly less than 1%, and other adverse
reactions less than 1%. All of these studies documented in the
literature combine to confirm the excellent safety profile of
Now, 20 years after this group of agents was initially
developed, it is appropriate to go back and look at the fundamental
safety basis for this group of agents. Two recent review articles
look at the overall safety for the approved gadolinium chelates.
The researchers found that the safety profiles for Magnevist,
Omniscan, and ProHance were comparable. They found a transient
elevation in serum iron with both Magnevist and Omniscan, believed
to be the result of some instability of the chelate and some
red-blood hemolysis with injection. One review also noted the
problem with higher osmolality contrast agents (specifically
Magnevist), namely that contrast extravasation can lead to tissue
necrosis. Table 3 presents a comparison of safety data from the
U.S. clinical trials of the four agents currently on the market.
The data confirm that these agents cannot be differentiated on the
basis of adverse reactions. For example, the incidence of nausea in
U.S. clinical trials was 1.9% for Magnevist; 1.6% for Omniscan;
3.2%, for OptiMARK; and 1.7% for ProHance.
In conclusion, it is clear from the literature that the safety
profiles of all the currently approved gadolinium chelates are
similar. Headache, nausea and hives remain the most commonly seen
reactions and are generally self-limiting. The headache reported in
clinical trials is thought to be due to the exam conditions rather
than to the agent itself. Severe, potentially life-threatening
reactions, although very rare, have been seen with all the
gadolinium chelates, and practitioners should be prepared to treat
them. Some differences do exist between the agents due to their
physicochemical properties. For example, a higher osmolar agent
such as Magnevist (1,960 mOsmol/kg) is more viscous (and therefore
harder to inject by hand) and may cause more local irritation upon
injection and more tissue damage in case of accidental
extravasation as compared to a lower osmolar agent such as ProHance
(630 mOsmol/kg). In most clinical settings, however, the safety
profile of the available gadolinium chelates is virtually