Abstract
Background and objectives: Recent studies strongly link nephrogenic systemic fibrosis to gadolinium administration for magnetic resonance imaging. In a recent advisory, the Food and Drug Administration stated that all gadolinium-containing chelates are potentially associated with nephrogenic systemic fibrosis; however, most reported cases are linked to gadodiamide (Omniscan) and gadopentetate dimeglumine (Magnevist). Given the severe consequences of nephrogenic systemic fibrosis, it is critical to define the risks associated with each gadolinium-containing chelate. The purpose of this study was to examine nephrogenic systemic fibrosis risk in a hemodialysis population exposed to gadoteridol (ProHance).
Design, setting, participants, & objectives: Appointment logs were used to generate a database of all long-term hemodialysis patients at the Dallas Veterans Affairs hospital since August 2001. These patients were then examined in the Veterans Affair's electronic medical record system for gadolinium exposure during magnetic resonance imaging from 2000 through 2007, a period during which gadoteridol was the sole contrast agent used.
Results: A total of 141 patients were identified with 198 gadoteridol exposures. No cases of nephrogenic systemic fibrosis were identified. The observed frequency of nephrogenic systemic fibrosis was compared with the expected frequency (2.4%) using one-way χ2 and binomial analysis, yielding a P < 0.05, indicating that the result was not explained by chance alone.
Conclusions: It is concluded that the risk for nephrogenic systemic fibrosis with gadoteridol in patients who are on long-term hemodialysis may be lower than with gadodiamide and gadopentetate dimeglumine.
Nephrogenic systemic fibrosis (NSF) is a devastating disorder characterized by progressive fibrosis in skin and multiple other organ systems. It occurs only in patients with reduced renal function. Recent studies have strongly linked NSF to gadolinium administration for magnetic resonance imaging (MRI) (1–6). In a recent advisory, the US Food and Drug Administration (FDA) stated that it views all five FDA-approved gadolinium-containing chelates (GCC) as potentially associated with NSF. This advisory carries the danger that clinicians will misinterpret this to mean that the NSF risk is equal with all five FDA-approved GCC; however, literature reports and information from the FDA's MedWatch suggest that there may be differing levels of NSF risk for GCC (7).
Gadolinium has seven unpaired electrons in its outer shell and as a result is strongly paramagnetic and an ideal contrast agent for MRI. Free gadolinium, however, is highly toxic and relatively insoluble in water and as a result must be chelated to be used in humans. Each of the five FDA-approved GCC contains a unique chelate, as well as differing amounts of free chelate. The chelates may be either ionic or nonionic and either linear or cyclic. In general, ionic chelates tend to bind gadolinium more avidly than nonionic chelates. In addition, cyclic chelates bind gadolinium more strongly than linear chelates. If the pathophysiologic mechanism of NSF is related to gadolinium release, deposition, or binding in tissues, then one can make inferences regarding the potential risk of each GCC given their thermodynamic stability constants, dissociation half-lives, and the relative amount of free chelate in each preparation (Table 1). Chelates that bind gadolinium more avidly as evidenced by a higher thermodynamic stability constant and longer dissociation half-life or that contain more excess free chelate to scavenge released gadolinium might be associated with a lower NSF risk. Gadodiamide (Omniscan, GE Healthcare, Waukesha, WI) contains a linear nonionic chelate and has a very short dissociation half life (30 s). Gadopentetate dimeglumine (Magnevist, Bayer HealthCare Pharmaceuticals, Montville, NJ) has a linear ionic chelate and as a result has an intermediate dissociation half-life (10 min). The chelating molecule in gadoteridol (ProHance, Bracco Diagnostics, Inc., Princeton, NJ) is cyclic and nonionic and has the longest dissociation half-life (3 h) of any of the FDA-approved GCC.
Gadolinium-containing chelates
In a recent study by Deo et al. (8) in a group of hemodialysis patients in Bridgeport, CT, who were exposed to gadodiamide and gadopentetate dimeglumine, it was estimated that the risk for NSF per exposure was 2.4%. In our institution, gadoteridol has been used exclusively in MRI since 2000. We hypothesized that given the structure of the chelating molecule, gadoteridol (ProHance) might be associated with a lower risk for NSF in chronic hemodialysis patients than previously reported with gadodiamide (Omniscan) and gadobenate dimeglumine (Magnevist).
Materials and Methods
Study Design
Appointment logs were used to generate a database of all long-term hemodialysis patients who were seen at the Dallas Veterans Affairs (VA) hospital since August 2001. These patients were examined using the Veterans Health Information Systems and Technology Architecture appointment manager system in Computerized Patient Record System for gadolinium exposure during MRI from 2000 through 2007, a period during which gadoteridol was the sole contrast agent used. Patients with <14 d of follow-up after exposure were excluded. A systematic search for NSF was carried out by examination of each patient's electronic medical record for all dermatology consults, all pathology reports, the problem list for any dermatologic condition, and all discharge summaries. In addition, a keyword search of every progress note using the term “nephrogenic” was carried out. During this period, all patient entries were made solely into the electronic medical record. For each patient, data were also collected for date of birth, gender, race, ethnicity, date of hemodialysis initiation, age at the time of exposure, type of study, and indication for the procedure. Most recent data from the FDA's MedWatch were also obtained via a freedom-of-information request. MedWatch data result from voluntary reporting by providers. It contains number and type of GCC exposures, demographic descriptors of the patient, and some information regarding other medications and associated medical conditions.
Statistical Analyses
The observed versus expected rates of NSF were compared using one-way χ2 with 1 df and binomial analysis. Standard descriptive statistics such as mean and SD were used to describe data collected when appropriate. The t test for two independent sample means was used to compare the means for the number of gadolinium exposures in MedWatch-reported cases of NSF with gadopentetate dimeglumine and gadodiamide.
Results
A total of 141 patients were identified with 198 gadolinium exposures. Demographic data for the 141 exposed patients are shown in Table 2. Our dialysis unit has a large percentage of black patients. The overrepresentation of black patients in the ESRD population is well documented and may be related to genetic factors (9). The corresponding racial and ethnic composition of Dallas County in the 2006 US census was 72.5% white alone, 37.7% Hispanic, and 21.1% black (10). Men make up a large fraction of patients as is expected within the VA. Currently, 8% of veterans who are cared for within the VA North Texas Health Care System are women. The average age at the time of each exposure was 61.8 ± 9.8 yr. The majority of patients received gadoteridol once. Two patients were exposed on five separate occasions, one during a 2-mo period and the other during a 5-mo period.
Demographics
There were 111 exposures from MRI examinations and 87 gadoteridol exposures with angiography or venography. The type of study performed for each exposure is delineated in Table 3.
Studies using gadoliniuma
Despite an extensive search of each patient's electronic medical record, no cases of NSF were detected. Deo et al. (8) found that the risk for NSF in hemodialysis patients in Bridgeport, CT, was 2.4% per gadolinium exposure. Patients who developed NSF in that study were exposed to gadodiamide (Omniscan) and gadopentetate dimeglumine (Magnevist). The observed frequency in our study was compared with the expected frequency (2.4%) using both one-way χ2 with 1 df and binomial analysis. Both tests yielded a P < 0.05, indicating that the result was not explained by chance alone.
Data were also obtained from the FDA's MedWatch through a freedom-of-information request. It should be cautioned that data from MedWatch are incomplete given that volunteer case reports are not subject to peer review and documentation is less stringent than in the published literature; as a result, conclusions drawn from MedWatch should be viewed with caution. Several interesting observations emerge from the individual reports as of October 23, 2007 (Table 4). The majority of reported associations are with gadodiamide (Omniscan), and in 246 of 283 gadodiamide-associated cases, gadodiamide was the only GCC administered. Gadopentetate dimeglumine (Magnevist) was the second most frequent GCC associated with NSF, and in 96 of 125 gadopentetate dimeglumine-associated cases, it was the sole GCC reported. When compared with other GCC-associated cases, for those that were associated with gadopentetate dimeglumine, the patient received that agent more frequently than gadodiamide (2.7 ± 2.43 versus 1.28 ± 1.07, respectively). This was statistically significant using the t test for unpaired samples. Although gadoversetamide (OptiMARK, Covidien, Hazelwood, MO) was less frequently associated, it is concerning that in eight of 20 reports, gadoversetamide was the sole GCC administered. Gadoteridol (ProHance) and gadobenate dimeglumine (MultiHance, Bracco Diagnostics, Inc., Princeton, NJ) were associated with nine and 10 cases, respectively. In only two of 10 cases associated with gadobenate dimeglumine was it the sole GCC administered. Of the nine cases of NSF associated with gadoteridol, in only one was it the sole GCC administered, and that patient received five doses during a 2-yr period (11).
FDA MedWatch reported casesa
Discussion
NSF is a multisystem disorder that may result in significant morbidity and mortality in patients with renal dysfunction. It is recommended that informed consent be obtained from patients with kidney disease before GCC administration (12); however, it is difficult for the clinician to counsel patients adequately on the risk for NSF given the currently available literature. The FDA in a recent advisory stated that all GCC carry the potential risk for association with NSF (13). Although this is true, there is a concern that clinicians will misinterpret this to mean that NSF risk with the five FDA-approved GCC is equal. As discussed here, on the basis of our own study and an analysis of MedWatch data, the risk for NSF with the five FDA-approved GCC is likely not equal.
The vast majority of reported NSF cases (93 of 117) are associated with gadodiamide, and the American College of Radiology recommends that gadodiamide not be used in patients with renal disease (12). Gadopentetate dimeglumine makes up most of the remainder of reported cases (18 of 117), and recently it was recommended by the UK Medicine and Healthcare Products Regulatory Agency that patients with kidney disease not be administered this agent as well (14). Manufacturers’ websites and MedWatch report cases of affected patients, but these data are difficult to interpret without knowledge of market share for each of the five FDA-approved GCC. To date, 408 cases of NSF associated with GCC have been reported to MedWatch. Of these 408, 353 were associated with reporting of only one gadolinium preparation and 55 occurred when the patient was reported to be exposed to multiple preparations. At least 29.4% of patients received more than one gadolinium exposure. This may be an underestimation as a result of the way in which cases are reported. In the study by Deo et al. (8), three patients developed NSF (two with gadodiamide and one with gadopentetate dimeglumine) after a total of 123 gadolinium exposures; however, the number of patients who were exposed to each individual agent was not reported.
One would predict that gadodiamide (Omniscan) and gadoversetamide (OptiMARK), given their linear, nonionic chelate and similar thermodynamic stability constants (16.8), would be most likely to release gadolinium from the chelate and, therefore, be associated with the highest NSF risk. Gadodiamide has a dissociation half-life of only 30 s. The published literature and MedWatch data support an association for gadodiamide; however, there are no reported cases in the literature for gadoversetamide, and only 20 cases were reported to MedWatch. It is interesting that a high percentage of these 20 cases occurred with reporting of gadoversetamide exposure alone (40%). The difference in the number of reported cases between these two GCC has several possible explanations, including but not limited to the following: (1) The increased amount of free chelate in gadoversetamide (28.4 versus 14 mg/ml with gadodiamide) scavenges free gadolinium released from the chelate and is associated with a lower NSF risk; or (2) NSF risk with these agents is similar but differences in case reports and MedWatch reporting are due to differences in market share. Although exact US and worldwide market share data are difficult to obtain, within the VA between July 1, 2005, and September 30, 2007, gadodiamide had a larger market share than gadoversetamide: 25.6 and 6.1%, respectively (15). As a surrogate for the US market share, we used the VA purchasing records for GCC. The VA does not have a national contract for the purchase of GCC, and each individual VA hospital determines which GCC it purchases similar to that of non-VA hospitals across the country.
Gadopentetate dimeglumine (Magnevist) and gadobenate dimeglumine (MultiHance) contain linear ionic chelates. They have higher thermodynamic stability constants (22.2 and 22.6, respectively) than either gadodiamide (Omniscan) or gadoversetamide (OptiMARK). The dissociation half-life of gadopentetate dimeglumine is 10 min, and that for gadobenate dimeglumine has not been reported. A total of 30.6% of NSF cases that were reported to MedWatch were associated with gadopentetate dimeglumine. In 97 of 125 MedWatch cases that were associated with gadopentetate dimeglumine, it was the sole gadolinium-containing chelate administered. In 54.6% of these 97 cases, the patient received multiple doses. The interval during which this occurred is unclear because of the incompleteness of MedWatch reporting, and MedWatch data may also underrepresent the number of patients who received multiple doses. At least one study has linked the size of the gadolinium dose administered to the relative risk for association with NSF (4). One can extrapolate that the number of exposures and the period of time during which they occur may also affect the association. Whether this is particularly true of cases that are associated with gadopentetate dimeglumine remains to be determined. As shown in Table 4, the number of gadopentetate dimeglumine exposures per patient in MedWatch cases was higher (2.7 ± 2.43) than with any of the other four FDA-approved GCC. Gadobenate dimeglumine is associated with no case reports of NSF, and only 10 cases were reported to MedWatch. In only two of these 10 cases was gadobenate dimeglumine the sole GCC administered. Whether these dramatic differences in NSF association are confounded by market share or result from differences in other physicochemical properties is unclear. Within the VA between July 1, 2005, and September 30, 2007, gadopentetate dimeglumine had a much larger market share than gadobenate dimeglumine (53.9 and 3.2%, respectively) (15). The company websites for each agent list the number of administered doses worldwide as greater than 80 million for gadopentetate dimeglumine and 3.5 million for gadobenate dimeglumine (16,17).
Gadoteridol (ProHance) contains a cyclic nonionic chelate. It has the highest thermodynamic stability constant (23.8) and the longest dissociation half-life (3 h) of the five FDA-approved GCC. One might, therefore, predict that given these properties, it would be associated with the lowest NSF risk. Gadoteridol is associated with the fewest number of reported MedWatch cases (nine). It is interesting that of these cases, only one occurred with gadoteridol administered alone, and that patient had six exposures during a 2-yr period. There are no case reports to date linking NSF to gadoteridol exposure. It is possible, although less likely, that the low number of MedWatch cases and lack of reported cases are related to market share. Within the VA from September 1, 2005, to September 30, 2007, gadoteridol use was 11.1% of all GCC. The manufacturer's website indicates that more than 8.5 million doses of gadoteridol have been administered worldwide (18). In our study, we did not detect a case of NSF with 198 exposures. Given that no cases were detected, an exact incidence rate cannot be calculated, but we found that the number of cases detected was statistically significantly different from that of Deo et al. (8).
Given the nature of the study, it is possible that milder presentations of NSF were missed. Whether milder phenotypes of NSF are present is unclear and was suggested by one group (19). Hopefully, biopsies of exposed patients with suspected disease can be compared with controls with exacerbating and remitting edema from intermittent volume overload to address this issue more definitively in hemodialysis patients. It is also possible that some full-blown cases of NSF may have been missed by clinicians. Alternatively, a less likely explanation is that other risk factors associated with NSF, such as edema, endothelial damage, and liver disease, were underrepresented in our inner-city, hospital-based dialysis unit.
Conclusions
The observed frequency of NSF in our study was compared with the expected frequency (2.4%) using both one-way χ2 and binomial analyses. Both tests yielded a P < 0.05, indicating that the result was not explained by chance alone. We conclude that NSF risk in patients who have ESRD and are on long-term hemodialysis may be lower with gadoteridol (ProHance) than with gadodiamide (Omniscan) and gadopentetate dimeglumine (Magnevist). Further studies will be required to confirm this finding and better delineate NSF risk with different gadolinium preparations.
Disclosures
None.
Acknowledgments
This work was supported by the University of Texas Southwestern Medical Center at Dallas O'Brien Kidney Research Core Center (P30DK079328).
Parts of these data were presented in abstract form at the annual meeting of the American Society of Nephrology; November 2 through 5, 2007; San Francisco, CA.
Footnotes
Published online ahead of print. Publication date available at www.cjasn.org.
See related editorial, “How Should Nephrologists Approach Gadolinium-based Contrast Imaging in Patients with Kidney Disease?” on pages 649–651.
- Received December 26, 2007.
- Accepted January 22, 2008.
- Copyright © 2008 by the American Society of Nephrology
In this issue
Jump to section
More in this TOC Section
Cited By...
- Nephrogenic Systemic Fibrosis: Review of 370 Biopsy-Confirmed Cases
- Nephrogenic Systemic Fibrosis: A Survey of Nephrologists' Perceptions and Practices
- Nephrogenic Systemic Fibrosis: Pathogenesis, Diagnosis, and Therapy
- Current Status of Gadolinium Toxicity in Patients with Kidney Disease
- Nephrogenic Systemic Fibrosis: Considerations for the Cardiologist