Christine U. Lee, Christina M. Wood, Gina K. Hesley, Nelson Leung, Mellena D. Bridges, Jeffrey T. Lund, Peter U. Lee, Mark R. Pittelkow. Large Sample of Nephrogenic Systemic Fibrosis Cases From a Single Institution. Arch Dermatol. 2009;145(10):1095–1102. doi:10.1001/archdermatol.2009.232
To estimate and stratify the risk of development of nephrogenic systemic fibrosis (NSF) in well-defined at-risk subpopulations from a large single institution, and to perform a single-institution case series study of patients with biopsy-proven NSF.
Retrospective cohort of patients exposed to gadolinium-based contrast agents (GBCAs) at a single institution during an 8-year period (January 1, 1999, to December 31, 2006), and a case series study of patients with biopsy-proven NSF.
A primary, secondary, and tertiary health care center that treated more than 2.2 million outpatients and had 135 000 hospital admissions in 2007.
A total of 94 917 patients exposed to GBCAs; patients at risk for NSF (3779 patients on hemodialysis, 1694 patients with renal transplants, and 717 patients with liver transplants, a well-defined subgroup that includes patients at risk for reduced renal function); and 61 patients with a clinical diagnosis of NSF.
Main Outcome Measure
Risk estimate for NSF.
The risk of development of NSF is 1.0% for patients who undergo hemodialysis (8 of 827), 0.8% for patients with renal transplantation (4 of 527), and 0% for patients with liver transplantation at our institution (0 of 327).
Despite the limitations, this study, which reviewed a large number of patients who underwent intravascular GBCA injections, demonstrates a 77-fold higher risk of NSF among patients who undergo hemodialysis and a 69-fold higher risk in patients with renal transplantation. This increased risk is thought to be associated with poor clearance of most GBCAs.
Nephrogenic systemic fibrosis (NSF) remains a pathophysiologic enigma despite more than a decade of investigation. Its impact, mostly owing to its association with gadolinium-based contrast agents (GBCAs), has been profound, especially among those patients considered at risk. Determination of the risk of development of NSF, in its most rigorous sense, has been confounded by limitations of access to medical records, incomplete documentation of lifetime GBCA exposure of a patient, and varied laboratory assessments of renal function before GBCA administration as well as confounding comorbidities near the time of GBCA exposure.1- 9 It has been well-documented that patients with a history of moderate-to-severe renal dysfunction, renal transplantation, hemodialysis, and hepatorenal syndrome are at increased risk.1,3,6,10- 12
Most investigatory efforts that focus on signs and symptoms as well as cross-sectional studies estimate risk that ranges from 1% to 7% among at-risk patients.1,6,11,13 Risk stratification has remained a challenging endeavor, primarily because of the imprecise understanding of the pathophysiologic mechanisms of NSF. This study is perhaps the largest database from a single institution; in 2007 an initial study from our institution reported lower risks for development of NSF, compared with other institutions, across several subpopulations.14 The purpose of this study is to describe the NSF experience at our institution. Specific objectives are 2-fold: to estimate the risk of development of NSF in at-risk subpopulations at our institution and to conduct a case series study of patients with biopsy-proven NSF.
A retrospective review was performed following approval by the Institutional Review Board and maintainance of HIPAA (Health Insurance Portability and Accountability Act) compliance. Preliminary retrospective review of patients seen during an 11⅓-year period (January 1, 1996, to May 31, 2007) resulted in clinical and diagnostic terms not entirely conclusive for a diagnosis of NSF, particularly because this entity was not described until 1997. For one of the earliest cases of NSF at our institution, the terms “NSF” or “nephrogenic fibrosing dermopathy” were not designated. Therefore, the study period was narrowed to an 8-year time interval. Beginning in 1999, clinical records at our institution were all electronic, which reduced the chances of missing a case of NSF.
We identified all patients exposed to GBCAs between January 1, 1999, and December 31, 2006, by the searching of all Current Procedural Terminology codes, indication notes, procedural notes, and operative notes for keywords that included all brand names of GBCAs used at our institution: gadolinium, contrast, Omniscan (GE Healthcare, Waukesha, Wisconsin), Magnevist (Bayer HealthCare Pharmaceuticals, Wayne, New Jersey), ProHance (Bracco Group Inc, Princeton, New Jersey) and MultiHance (Bracco Group). This resulted in evidence of 149 460 doses of GBCA administered to 94 917 patients during the study period (Table 1).
Every attempt to retrospectively identify all patients with renal disease before GBCA exposure failed. The values of variables needed to calculate estimated glomerular filtration rates (eGFRs) such as creatinine, age at exposure, and race were confounded by nonuniform clinical situations and ordering protocols. The determination of renal function in patients with end-stage liver disease, included in the recommendation by the United States Food and Drug Administration (FDA),15 is often difficult. Nephrogenic systemic fibrosis has been described in liver transplantation patients16, and the FDA has issued similar cautionary advice in the perioperative liver transplantation period. Patients with serum creatinine levels within the reference range can have a markedly reduced glomerular filtration rate.17 Cirrhosis is also a risk factor for reduced renal function. As such, in addition to the well-defined subpopulation of hemodialysis and renal transplantation cases from 1999 to 2006, we included the well-defined liver transplantation cohort, which includes those patients with cirrhosis who may be at risk for development of NSF. Details about these subpopulations are provided in Table 2.
A medical index retrieval specialist electronically searched millions of clinical records, the coded medical index database, and our institutional Life Sciences System text tool (Mayo Clinic, Rochester, Minnesota) for the terms “scleroderma,” “morphea,” “nephrogenic fibrosing dermopathy,” “fibrosing dermopathy,” “NFD,” “nephrogenic systemic fibrosis,” and “NSF” within patient records. Exclusion of the search term “scleroderma” resulted in 788 patients, of whom 143 gave research authorization. For these 143 patients, medical records, including dermatopathologic data, were reviewed. All patients with NSF seen by other physicians who practice subspecialties without departmental NSF databases, such as rheumatology, were captured in this way. By the pooling of these data with data from known interdepartmental NSF databases (Nephrology, Dermatology, and Radiology), we identified 37 patients diagnosed as having histologically proven NSF during the 8-year study period (Table 3). Of the 37, 11 had received GBCAs at our institution before histologic diagnosis. The remaining 26 either did not receive GBCAs at our institution before histologic diagnosis or had uncertain GBCA exposure histories. Medical records for the 11 patients with NSF who had antecedent exposure to GBCAs were reviewed with particular attention to renal function data. Because eGFRs are invalid among patients who undergo dialysis, these patients were assigned an eGFR value of “0.”
After Institutional Review Board approval, we contacted 8 of 26 patients with NSF who had uncertain GBCA exposure histories to request lifetime medical records and details of GBCA exposure at other health care facilities; we specifically requested any magnetic resonance imaging records and data with regard to interventional procedures performed with contrast, as well as dialysis, laboratory, internal medicine, and, particularly, nephrology records. Seven patients responded; 5 assented, and 2 declined. Review of extensive records from these 5 patients yielded disappointingly minimal, if any, additional information about prior gadolinium exposure. This lack was attributed to deficiencies in record keeping, and, simply, to the fact that medical records were not provided.
To clinically characterize our NSF cohort, we not only discarded statistical time constraints and used a wider study period that encompassed 9 years, 5 months, from January 1, 1998, to May 31, 2007, but also included NSF databases from off-campus institutional sister sites. A total of 61 NSF cases, 57 with histologic evidence, were documented during this period at our institution. We retrospectively reviewed records from 57 patients with histologically proven NSF, with focus primarily on age, weight, renal status, comorbidities, onset of skin symptoms, date of pathological diagnosis, and GBCA specifics (eg, brand, route, dosage, frequency of exposure). The renal status of each patient was determined from medical records and eGFRs by means of the serum creatinine level documented nearest the time of GBCA exposure.18
The 94 917 patients exposed to GBCAs during the 8-year study period (Table 1) had between 1 and 36 exposures. Median age at first exposure was 55 years (interquartile range, 41-69 years), and 52.4% of patients were women. Of 149 460 doses, gadodiamide (Omniscan) was used 69.3% of the time. A total of 99.4% of GBCA doses were administered for magnetic resonance imaging studies, and less than 1% was administered for interventional radiologic and cardiologic procedures. Five of 11 patients with NSF were already undergoing dialysis at the time of GBCA exposure; 3 were not undergoing hemodialysis at the time of GBCA exposure but were at some point afterward, and 3 received renal transplants. Table 4 provides additional details about the 8 patients with NSF who were undergoing dialysis and received a GBCA before histological diagnosis.
For anyone who received GBCAs at our institution, the risk of development of histologically proven NSF from 1999 though 2006, defined as the number of patients who had histologically proven NSF after exposure to GBCAs divided by the number of patients administered a GBCA, was 0.012% (95% binomial confidence interval, 0.001%-0.021%). If the order of events was not specified (ie, GBCA exposure could occur before as well as after the diagnosis of NSF), then 13 of 95 220 at-risk patients (0.01%) exposed to GBCAs at our institution also had biopsy-proven NSF. Similarly calculated risks were 1.0% for patients who underwent hemodialysis, 0.8% for patients who underwent renal transplant, and 0.0% for patients who underwent liver transplant because none had received GBCAs at our institution before histologic diagnosis (Table 5).
The number of path-proven NSF cases per year was as follows: 1998, 1 case; 2000, 2 cases; 2002, 5 cases; 2003, 9 cases; 2004, 10 cases; 2005, 7 cases; 2006, 15 cases; first 5 months of 2007, 8 cases. Several past cases were reviewed because data indicated histologic similarity to NSF; 2 cases, such as the one in 1998, were reclassified from a diagnosis of pseudoscleroderma verus scleromyxedema and septal liposclerosis with minimal inflammation to NSF. The general increasing trend to diagnose NSF likely reflects its evolution as a disease entity that gained more diagnostic reliability in recent years, although increasing use of GBCAs, first approved by the FDA in 1988, may have also played a role.
The median age at the time of pathologic diagnosis was 64.5 years (age range, 19-83 years), and 27 of 57 (47.4%) patients were women. The median age in this study is slightly older than mean ages reported in other studies, which is possibly related to the fact that it took longer to make a pathologic diagnosis in the older, usually sicker, population. In addition, multiple exposures to GBCAs, if relevant, would be more likely among older patients.
Details about GBCA exposure and renal status of patients with NSF are shown in Table 6. Of 5 patients with an eGFR of approximately 30, 2 had undergone liver transplantation and the third was in a posttraumatic state at the time of NSF diagnosis. One of the patients who received a liver transplant had an eGFR of 30 when GBCA was administered, 6 months before transplantation. Thus, GBCA was administered twice: once 12 days posttransplantation (eGFR, 24) and again 6 months posttransplantation (eGFRs, 42 and 50 measured twice during the medical workup). The other patient who underwent liver transplantation was exposed to GBCAs twice: once 2 months before transplantation (eGFRs, 18 and 24) and again during the peritransplant period (eGFRs, 31 and 47). The patient in a posttraumatic state with resultant paraplegia had a history of diabetes and underwent 2 studies with the use of GBCAs during a 6-day period after the traumatic incident: once with an eGFR of less than 30 and again with an eGFR more than 30.
The mean cumulative volume of GBCAs administered before histologic diagnosis was 54 mL in the NSF population, but in the at-risk populations, the mean volumes were smaller: 42 mL in the hemodialysis group, 45 mL in the renal transplantation group, and 36 mL in the liver transplantation group. At our institution, clinical scenarios near the time of GBCA exposure reflected a patient cohort that was generally unwell; many patients required hospitalization, with comorbidities summarized in Table 7.
The pathophysiologic mechanism of NSF and its association with GBCAs and renal disease remains uncertain. In the context of such uncertainty, medicolegal implications continue to place an enormous burden on medical practice and patient safety. Governing agencies worldwide have made recommendations informed by risk assessment to health care professionals in an effort to alleviate some of this burden.15,20 However, the determination of risk has been anything but straightforward, particularly in large retrospective epidemiologic studies such as this one. The search for factors with that may play significant roles in the pathophysiologic mechanism of NSF has been limited and compounds the frustrations of interpretation. Inconsistencies in contrast-enhanced magnetic resonance imaging records, which have been a little more than 20 years in the making, are disappointing, if not embarrassing. Data from studies that use gadobenate dimeglumine (MultiHance) and gadoteridol (ProHance) were confounded by prior administrations of either gadodiamide (Omniscan) or gadopentetate dimeglumine (Magnevist).
Although this study cannot provide definitive clinical recommendations or absolute contraindications to GBCA administration, it does offer some additional insight into the screening of patients. In patients with liver failure, renal function can be overestimated when only eGFRs are used as measurement, as demonstrated in the liver failure and transplantation patients in our cohort. However, this factor will likely not be considered by the allied health personnel who generally perform the screening and, thus, will need to be reconciled by physicians. Additionally, the positive association between cumulative dose of GBCAs and dosing events may require further investigation.21 At least for the NSF cohort in this study, less GBCAs may be required to initiate the development of NSF in patients who undergo hemodialysis, renal transplantation, and liver transplantation than in those patients with moderate-to-severe renal insufficiency. The brand, dosage, route, and cumulative dose of GBCAs over time remain critical considerations in day-to-day clinical practice. Definitive conclusions with regard to dose dependency and frequency of GBCA exposure cannot be extracted from this study. Perhaps determination of a minimum threshold dose for GBCAs will minimize or exclude any adverse effect.
Patients with acute renal insufficiency are classified as at risk by the FDA. Further stratification of patients by acuity and chronicity of renal failure was not possible in this study for the same reasons that the population with renal disease could not be obtained. During data review, a patient who subsequently developed NSF was hospitalized and being treated by one of us (N.L.). This patient developed acute renal failure and subsequently received 3 doses of a GBCA before skin symptoms arose and while renal function continued to decline. For this patient with acute renal failure, it is difficult to determine whether the first GBCA exposure played a role in the eventual development of NSF, relative to the latter 2 exposures.
Nephrogenic systemic fibrosis has been reported in children.22 In our data, 5.3% (n = 5056) of all first GBCA exposures were in patients younger than 18 years. The fact that no pediatric patients with NSF were identified is likely owing, in part, to the very few reported pediatric NSF cases.23,24 All our patients with NSF have been associated with GBCAs by thorough ascertainment. Although we have not performed any survival analysis, 15 deaths associated with NSF have occurred.
In our study, no patients with histologically proven NSF had normal or mildly decreased renal function (eGFR, >30 mL/min/1.73m2), which agrees with current consensus that NSF is a disease that occurs only among patients with moderately to severely decreased renal function. Renal disease has become the accepted sine qua non of NSF, and, because most GBCAs (Magnevist, Omniscan, OptiMARK, and ProHance) are nearly entirely renally cleared, it is probably a true asssertion. However, there has been increasing use and marketing of newer generation GBCAs with some measure of biliary clearance, such as gadoxetate (Eovist; Bayer HealthCare; approximately 50% hepatobiliary clearance) and gadobenate dimeglumine (MultiHance; Bracco Group; <5% hepatobiliary clearance). This certainly raises the question of whether poor biliary clearance would increase the risk of NSF in the absence of renal dysfunction. With that in mind, it is perhaps more accurate, to state that poor clearance of GBCA is a requirement for the development of NSF.
Despite our best efforts to overcome the limitations inherent in the database, we were unable to retrospectively and accurately discern the population with renal disease from which clinically significant statistics could have been derived. The tertiary referral nature of our large institution is such that many patients receive initial care there but have follow-up care elsewhere. Determination of exactly how many patients return for follow-up care at our institution was, surprisingly, a difficult process plagued with its own sources of error. To circumvent this, we looked specifically at the cohort given a GBCA at our institution (Table 2). Among 527 patients who underwent renal transplantation, patients returned for follow-up care at a median of 2.9 years, and 464 (88.0%) had 1 day or more of follow-up. Among 827 patients who underwent hemodialysis, patients returned for follow-up at a median of 1.2 years, and 522 (63.1%) had 1 day or more of follow-up. Among 327 patients who underwent liver transplantation, the median follow-up time was 2.0 years, and 251 (76.8%) had 1 day or more of follow-up. Although the tertiary nature of our large institution contributes to a different epidemiologic experience, it is similar to the combined experiences from 2 large medical centers25 and from 4 American universities,25,26 published in 2008.
Sources of error are primarily related to the nature of the database. Some patients exposed to gadolinium at our institution may have developed NSF and been seen elsewhere without any follow-up or annotation in the medical records at our institution. Therefore, they may have been included in the denominator for calculation of risk but were not included in the numerator. In this study, the date of histologic diagnosis was used, which excluded patients who had a strong clinical diagnosis of NSF but no histologic confirmation, again possibly contributing to an underestimation of risk. The low risk for NSF in the liver transplantation cohort is attributed, in part, to the small cohort size. The practice tendencies of our hepatologists are to use ultrasonography rather than magnetic resonance imaging in the peritransplant period. Additionally, many of our patients who underwent liver transplantation are those with hepatocellular carcinoma who have been given priority and who are typically not as sick as other patients who await transplantation. For the hemodialysis population, differences in dialysis protocols could also have contributed to an underestimation of risk.
The limitations of documentation were quite frustrating. For example, administered GBCAs were usually documented in volume rather than dose (millimoles per kilogram). Subsequent review of medical records showed that patient weights were not always documented near the time of GBCA exposure, which lead to inaccuracies in determination of the dose admininstered. Renal status, determined by creatinine level, was not always obtained.
Despite the limitations, this study demonstrates that no patients with NSF had creatinine levels within the reference range, which is in keeping with published consensus that NSF develops only in the setting of renal disease. We also found a 77-fold and 69-fold higher risk of NSF among the patients who underwent hemodialysis (1.0%) and renal transplantation (0.8%), respectively. Those GBCAs for which manufacturers tout biliary clearance remind us that it is, in fact, poor clearance, not just renal clearance, that plays a role in the development of NSF. The GBCAs add important diagnostic information to patient examinations, and one possibility to be considered is whether guidelines similar to those that form the basis of the ALARA (as low as reasonably achievable) principle with regard to limitation of radiation exposure should be implemented for GBCA administration for all patients.
Correspondence: Christine U. Lee, MD, PhD, Department of Radiology, Mayo Clinic, 200 First St SW, Rochester, MN 55905 (firstname.lastname@example.org).
Accepted for Publication: June 2, 2009.
Author Contributions: All authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: C. U. Lee, Wood, Hesley, Leung, and Pittelkow. Acquisition of data: C. U. Lee, Wood, Hesley, Leung, Bridges, Lund, P. U. Lee, and Pittelkow. Analysis and interpretation of data: C. U. Lee, Wood, Hesley, Leung, Bridges, Lund, P. U. Lee, and Pittelkow. Drafting of the manuscript: C. U. Lee, Wood, Hesley, Leung, Bridges, Lund, P. U. Lee, and Pittelkow. Critical revision of the manuscript for important intellectual content: C. U. Lee, Wood, Hesley, Leung, Bridges, Lund, P. U. Lee, and Pittelkow. Statistical analysis: Wood. Obtained funding: C. U. Lee and Hesley. Administrative, technical, or material support: C. U. Lee, Wood, Hesley, Leung, Bridges, Lund, P. U. Lee, and Pittelkow. Study supervision: C. U. Lee, Hesley, Leung, and Pittelkow.
Financial Disclosure: None reported.
Funding/Support: This study was supported by the Radiology 10-10-10-10 Award from the Department of Radiology, Mayo Clinic.
Role of the Sponsor: The sponsor had no role in the design and conduct of the study; in the collection, analysis, and interpretation of data; or in the preparation, review, or approval of the manuscript.