Volume 5, Issue 1 , Pages 23-28, January 2008
Nephrogenic Systemic Fibrosis: An Overview
Article Outline
- Abstract
- Introduction
- Patients at Risk
- Clinical Presentation
- Early Epidemiologic Investigations
- “Circulating Fibrocytes”
- Clues to the Cause
- What Is Gadolinium?
- Recent Citations
- Tissue Evidence
- Preliminary Animal Work
- Summary Opinion
- References
- Copyright
In 1997, a new fibrosing skin disorder became apparent among patients with renal disease. Nephrogenic systemic fibrosis, as this disorder has come to be known, has recently been tied to the administration of contrast agents containing gadolinium. This overview examines the discovery and elucidation of the major research accomplishments to date. It was presented as the keynote address at the First Annual Scientific Symposium on Nephrogenic Systemic Fibrosis and MRI Contrast, conducted at Yale University on May 4, 2007.
Key Words: Nephrogenic systemic fibrosis, nephrogenic fibrosing dermopathy, MRI contrast, MRI safety, pathology, epidemiology
The pure and simple truth is rarely pure and never simple.
—Oscar Wilde
Introduction
In 1997, Philip LeBoit, MD, a world-renowned dermatopathologist at University of California, San Francisco, received, in consultation, a number of pathology slides from the Sharp Medical Center in San Diego, California. These slides represented biopsy material from patients who had developed unexplained skin thickening. They were all renal dialysis patients in whom transplantation had been unsuccessful. The histopathologist in San Diego felt that the biopsies resembled a rare entity known as scleromyxedema. Because of this unusual focus among their dialysis population, they referred the biopsies to Dr LeBoit for evaluation. He reviewed the material and concurred: it did look like scleromyxedema. He deemed it a “scleromyxedema-like” disorder because the other clinical features of scleromyxedema were not present in these patients.
At that time, the epicenter for this disorder was in southern California. However, by 1999, the cases had become more widespread. There were cases in Ann Arbor, Michigan, a case in Mississippi, and several cases in northern California. The pattern seemed to be that as more physicians became familiar with nephrogenic systemic fibrosis (NSF) at their individual centers, they were able to recognize additional cases within their patient group. A recognition bias seemed to be occurring. But there was also concern that something peculiar about those individual centers was triggering NSF in these patients.
In 2000, our investigative team refined the disease definition enough to present a letter to The Lancet [1] warning that this disease seemed to be new, that it was appearing in a clinically definable group, and that we were seeking additional cases to try to more completely characterize the process.
Patients at Risk
In our earliest cases, transplantation was the factor that was common to all NSF patients. But as additional cases came to light, we began seeing patients who had never undergone transplantation. Most of those patients were on hemodialysis. Occasionally, we were seeing patients who had renal disease but had been on neither hemodialysis nor peritoneal dialysis. Initially, we thought that transplantation was related to the onset of the disease. Later, we thought that perhaps some dialysis membranes were related to the disease. In fact, we kept backing away from these hypotheses, ultimately concluding that the “at-risk” group could be defined as patients with renal disease. To date, I have not seen a convincing case of NSF that has arisen in a patient without renal disease.
We published our investigation of “nephrogenic fibrosing dermopathy” (the term we were using at the time), codifying the laboratory characteristics, the pathologic characteristics, and the epidemiologic findings [2]. In that initial group, there were 14 patients in 4 different US states. The age range was 31 to 74 years, with a male-to-female ratio of 9:5. In our most recent experience, we found that the age range spanned 8 to 86 years, with a male-to-female ratio of approximately 1:1. Many more patients, from diverse backgrounds, are now represented in the International Nephrogenic Systemic Fibrosis Registry at Yale University [3]. Native Americans and Pacific Islanders are represented, in addition to larger race groups. There is no clear race predilection.
In 2001, cases of NSF were confined to North America, but other places in the world have begun to deal with this emerging condition. At the time, we saw patients who had either transient, acute, or chronic renal disease. Some had been dialyzed, but some had not. They had numerous underlying renal conditions. It did not seem that the cause of the renal disease was important, but simply that the patients had renal disease [4].
Clinical Presentation
The clinical presentation of NSF commonly consists of skin papules and coalescing plaques (Figure 1), localized to the extremities and torso, creating an appearance reminiscing of the skin of an orange (known by the French term peau d’orange) and a wooden consistency [2]. Some patients manifest irregular erythema or brawny discoloration [5]. Disabling contractures of numerous joints may occur (Figure 2). Eye findings are seen in a large proportion of NSF patients [6]. These consist of plaques of whitish-yellow coloration with telangiectatic vessels, located on the sclerae (Figure 3).
Fig 1.
Skin findings: brawny to erythematous papules and plaques with a “peau d’orange” texture.
Fig 2.
Hand contractures in nephrogenic systemic fibrosis.
Fig 3.
Ocular findings: white to yellow scleral plaques with variable vascular ectasia.
A biopsy taken from skin from an involved cutaneous site reveals a thickened dermis containing more collagen than is normally seen [2, 5]. Collagen, mucin, and elastic can be deposited in increased amounts in patients with NSF. There is fibrosis of the septa within the subcutaneous fat. The septa may be 20 or more times the normal width. On higher magnification, there are increased numbers of bland spindle to epithelioid cells. They are not malignant. They are not mitotically active. When these are examined by immunohistochemistry scientists, long cellular processes marked by CD34 (a marker of vessels and stem cells) are seen. At low magnification, there is often a complex network of these processes throughout the dermis, extending down into the subcutaneous fat. If the sample is stained with procollagen I (a marker indicating collagen type I production), these cells turn out to be the same cells that stained with CD34 (although we did not recognize this in our initial description) [7]. Factor XIIIA–positive stellate cells are also expressed in increased numbers. CD68-positive cells (a marker of histiocytes) are also increased [2]. All of these cells are typically the background players in normal dermis, and they are all very much “activated” in NSF (Figure 4).
Fig 4.
Histologic composite: (A) Scanning magnification showing dense collagen throughout the dermis (pink). (B) High magnification showing thick (dark pink) collagen bundles with surrounding clefts and lighter pink elastic fibers. Numerous spindle-cell nuclei are seen. (C) CD34 stain highlighting (brown) a dense network of spindle-cell processes. (D) Procollagen I stain among the same spindle-cell population. The dual positivity of CD34 and procollagen I identifies these cells as “circulating” fibrocytes.
Early Epidemiologic Investigations
Shortly after we had characterized the clinical and pathologic aspects of NSF, the Centers for Disease Control and Prevention [8], which had been investigating the same group of patients in San Diego, published in the Morbidity and Mortality Weekly Report a paper titled “Fibrosing Skin Condition Among Patients With Renal Disease,” reporting a case-control study matching 3 controls per patient with NSF. The Centers for Disease Control and Prevention was not able to find any medication, toxin, infectious agent, or therapeutic technique emerging as a clear cause of NSF. Patients with NSF were more likely to have poor renal function after transplantation and more complicated clinical courses.
By 2003, NSF was more widespread and more widely recognized. As additional anecdotal reports and small case series became apparent, two articles appeared in the Journal of the Academy of Dermatology. One described several patients with NSF who had the scleral plaques described earlier [6]. The authors recognized these in all of their patients, and they pointed out that this was probably a manifestation of the disease. In the same journal issue, a second article described 4 of 4 patients with NSF who had demonstrable hypercoagulability [5]. I had also noticed that numerous patients seemed to have a thrombotic tendency [4]. I thought that it might be related to the onset of the disease, but I was unsure of the mechanism. All of the patients reported in these two studies had positive anticardiolipin antibodies, and one had NSF lesions before the onset of dialysis, illustrating that dialysis was not causing the disease.
In the summer of 2003, two events occurred that made us think differently about NSF. One was the publication of an article that described the first autopsy of a patient with NSF [9]. This patient also had a known history of anti–thrombin III and factor II deficiency and was also therefore hypercoagulable. The patient was so ill that he elected to discontinue dialysis because of his morbidity. This is not an unusual occurrence in patients with NSF. Nephrologists need to be aware that patients often feel this way. The autopsy findings described (besides cutaneous fibrosis) proximal esophageal fibrosis and diaphragmatic and psoas muscle fibrosis. This paper illustrated that NSF was more than just a skin disease; it seemed to be a systemic disorder.
“Circulating Fibrocytes”
Also in 2003, Richard Bucala, MD, hypothesized that the CD34-positive cells of NSF might in fact be circulating fibrocytes [7]. We investigated this possibility by performing a dual immunohistochemical stain, and we confirmed Dr Bucala’s suspicions. The simultaneous expression of two markers (CD34 and procollagen I) in these cells proved that these were in fact “circulating” fibrocytes. In the tissue of patients with NSF, they are not truly “circulating” anymore. When circulating, however, these bone marrow–derived cells uniquely coexpress these two antigens, and our belief is that they emerge from the circulation and cause the fibrosis of NSF. We published the findings of the dual positivity and our theory that these were in fact circulating fibrocytes in the American Journal of Dermatopathology in August 2003 [7]. In that article, we also included the logical basis for this pronouncement, which drew from more than just the immunohistochemical staining results I outlined above.
Later, in an article in Current Opinion in Rheumatology [4], I made a synopsis of all I had learned from the registry after 6 years of study. One of the things I discussed was the involvement of skeletal muscle in NSF. Skeletal muscle involvement, when observed, was noted immediately deep to involved skin. It seemed to be a direct contiguous extension of the disease from the skin down through the fascia to include the underlying skeletal muscle (Figure 5). With all of the additional proof that NSF was a systemic process, we elected to change the name from nephrogenic fibrosing dermopathy to NSF, because the new term was more reflective of our understanding of the disease at that time. For a while, there was overlap of the two terms, and this was intentional, because we did not want readers to believe that there were two distinctly different diseases. The terms describe (as they always have) the same disorder.
Fig 5.
Skeletal muscle involvement: normal skeletal fibers (left) and degenerating “beanbag”-like fibers (right).
Clues to the Cause
By 2004, there were many more reported cases, even some at small outlying clinics in sparsely populated areas. The geographical pattern of NSF emergence seemed to suggest that the largest population centers were seeing the most cases. We did not know whether this was because the physicians in those places were involved at academic centers and were therefore more aware of the disease (a recognition bias) or whether a large population was required to “see” the small numbers of patients who develop this disease. Perhaps, we speculated, there were techniques in larger cities that were not practiced in smaller communities. In 2006, we also learned of cases in other parts of the world.
In the article in Current Opinion in Rheumatology, I also explored the idea of thrombosis and surgery, because this association was coming up as a recurring theme in our registry [4]. Many of the patients who had thrombosis had underlying thrombotic tendencies. Sometimes they had not thrombosed, but when their physicians were asked to evaluate the patients for hypercoagulability, many discovered underlying thrombotic tendencies of which they were previously unaware. Additionally, many NSF patients had undergone recent vascular surgery. In some, NSF seemed to closely follow the vascular surgery, suggesting that events surrounding the surgery itself could be triggering this disease.
Again, a confluence of ideas occurred when Dr Grobner [10] published his article online in January 2006. He described gadolinium use in a population of 9 renal patients in Austria. Five of the 9 patients developed NSF. This was completely understandable to me, because I was independently investigating the thrombosis and vascular surgery associations, and I was finding that many of these patients had undergone a variety of types of angiographic procedures, and many, if not all, had received gadolinium. I was investigating independently the idea that gadolinium might be causing NSF. Dr Grobner presented his results first, and I completely concurred with his findings.
It was a startling coincidence and definitely worthy of further investigation. All of Dr Grobner’s patients had been histologically verified, which means they had all been biopsied, and the biopsies had been deemed consistent with NSF. All had received Omniscan (gadodiamide; GE Healthcare, Milwaukee, Wisconsin) [11]. There was a 2-week to 4-week delay between exposure and the development of symptoms of NSF. All patients had undergone magnetic resonance angiography. All had concurrent acidosis. So, the implication was that perhaps acidosis, in combination with gadolinium, was somehow causing NSF.
What Is Gadolinium?
Gadolinium is an element in the lanthanide series. Gadolinium is nonradioactive. It is paramagnetic, which means that it acquires unique physical characteristics when it is exposed to a magnetic field. Gadolinium is not “magnetic.” It is one of the most plentiful rare earth elements and is used in the manufacture of video recording heads, television screens, microwave ovens, and nuclear reactors. In its nonchelated (natural) form, it is considered highly toxic. Among its other effects, it is a well-known inorganic calcium channel blocker. Gadolinium is a component of magnetic resonance imaging contrast agents in use the world over. There are 5 gadolinium-based contrast agents currently approved for use in the United States: Magnevist (gadopentetate dimeglumine; Bayer Schering Pharma AG, Berlin, Germany), ProHance (gadoteridol; Bracco Diagnostics, Inc., Princeton, New Jersey), Omniscan, OptiMARK (gadoversetamide; Mallinckrodt, Inc., Hazelwood, Missouri), and MultiHance (gadobenate dimeglumine; Bracco Diagnostics, Inc.) (listed here in order of their introduction to the general public from 1998 through 2004). Currently, Magnevist and Omniscan are the subjects of the most intense investigation because they seem to be associated most frequently with NSF [3]. However, this association is complicated by the fact that they are also the two most widely used agents.
All gadolinium-based contrast agents are renally eliminated, and therefore, any patients who have nonfunctioning kidneys will be at risk for developing high levels of gadolinium in their bodies after exposure. Here I focus on Omniscan. In a patient with normal renal function, its half-life is 1.3 ± 0.25 hours. Omniscan is excreted very quickly, with no change in underlying renal function. The half-life of this agent in patients with end-stage renal disease is 32.3 ± 22.9 hours [12]. One hemodialysis session was able to achieve 65% elimination. Another paper examined a sequence of 3 hemodialysis sessions; in the first session, 78% was removed, then 96%, then 99%. One would have to undergo at least 3 hemodialysis sessions to result in 99% elimination of the agent [13].
In patients receiving peritoneal dialysis, one 22-day session resulted in only 69% elimination. These patients are therefore even worse off in terms of gadolinium elimination. The half-life in patients with peritoneal dialysis is 2.7 hours. Even if there is some minor residual renal function, the kidneys are not further affected by this agent.
How long does gadolinium persist in the body? In one investigation, using one animal (a dog with normal renal function), injected gadolinium was detected at 2 ppm in the liver up to 2 weeks after administration [14]. Therefore, even in a dog with normal renal function, there was a trace amount of gadolinium identifiable in the liver 2 weeks later. The authors of this particular study were not sure what level of gadolinium in tissues would be considered toxic, so they did not know how to interpret this finding. The Omniscan package insert (and package inserts from other companies) contains the following warning: “Pharmacokinetic and pharmacodynamic studies have not been systematically conducted to determine the optimal dose and optimal imaging time in patients with abnormal renal function or renal failure, or in the elderly or in pediatric patients with immature renal function.” In the package insert itself, it says that in renal disease, gadolinium could be potentially dangerous, or at least suggests that certain “at-risk” populations should be dealt with differently than the normal population.
Recent Citations
An article by Marckmann et al [15] provided additional proof for the suspected causative role of Omniscan in NSF. Those authors examined histologically verified cases. Their sample consisted of 380 renal patients who were imaged. Thirteen developed NSF, an incidence of 3.4%. All had received Omniscan. There was a 2-day to 75-day lag between exposure and the onset of the disease, and the authors did not find acidosis to be significant.
An article in Pediatric Nephrology in September 2006 described two patients with histologically verified NSF [16]. Both had received gadolinium-based contrast agents, although the brand was not indicated. Both had undergone magnetic resonance angiography. One patient had undergone a single magnetic resonance angiographic study. The second patient had a history of 13 magnetic resonance angiographic studies administered over a period of 4 years. This patient died 1 month after the last exposure to the agent. A time lag of at least 2 months between the exposure and the development of NSF was noted, and in this article, both patients were acidotic.
Another recent article in Investigative Radiology [17] examined 6 histologically verified cases of NSF. All the patients had received Omniscan. Two had undergone magnetic resonance imaging. Two had undergone magnetic resonance angiography. Two had undergone magnetic resonance venography. The onset of NSF was between 19 days and 2 months after exposure. Five of the 6 patients were acidotic.
In the guidelines of the European Society of Urogenital Radiology, Dr Thomsen stated that “gadodiamide [the generic name for Omniscan] should not be administered to patients with reduced kidney function or to those on dialysis” [18].
Tissue Evidence
We detected gadolinium in the tissue of patients with NSF [19]. We took 7 histologically verified cases of NSF from our registry. Four of these showed gadolinium in the tissue using electron microscopy with electron-dispersive spectroscopy. One patient (a negative control presented blinded to the investigators) had no demonstrable gadolinium. Patients had been exposed anywhere between 4 and 11 months before the biopsy, indicating that the gadolinium was clearly in the tissue for as long as 11 months since exposure. Once again, this proves that gadolinium hangs around a lot longer than previously thought in patients with renal disease. This deposited gadolinium seemed to be present in macrophages, as if it had been phagocytosed.
Preliminary Animal Work
Investigators from Bayer Schering Pharma AG performed animal injection studies [20]. They injected a large number of rats with a variety of agents. One was Bayer Schering’s own agent, Magnevist, and another was Omniscan. Some animals were injected with the active ingredient in Omniscan, without the extra ligand, to try to understand if the ligand molecule was protective against NSF. The most severely affected rat had been injected with Omniscan without the extra ligand (Omniscan is routinely packaged with excess ligand to provide shelf stability in the bottle). The clinical lesions of the animal were identical to the clinical lesions of human NSF.
Histologically, there were also similarities. Tissue from Magnevist-injected animals exhibited about twice as many spindle-cell nuclei as saline controls, but the animals did not exhibit the florid clinical lesions of fully developed NSF. These data are preliminary, because the images represent one animal in a series of many animals. The preliminary data, however, do suggest a difference between the saline-injected animals and the Magnevist-injected animals. The Omniscan-injected animals showed similar histology to fully developed human NSF. Those animals injected with Omniscan without extra ligand were most severely affected and had to be terminated before the study reached completion [20].
In summary, all of the animals used in this pilot study had normal renal function. To achieve the high tissue levels of gadolinium seen in patients who have abnormal renal function, investigators injected the rats with very high doses of gadolinium (25 times the proportional dose that a normal human would receive). In addition, they gave zinc supplementation to try to answer questions about induced zinc deficiency in gadolinium-exposed tissues. They compared saline, Magnevist, Omniscan, and gadodiamide (the generic name of Omniscan without the excess ligand). The saline and the Magnevist-treated animals were clinically indistinguishable. The animals did not lose fur. They did not have clinical lesions at 25 days, although the histologic changes in Magnevist (as noted above) were probably abnormal [20].
Comparing the animals who were obviously involved, gadodiamide (Omniscan without excess ligand) was much more severely involved than Omniscan, indicating that perhaps the chelator molecule is in some way protective against the development of this disease. The histology was completely compatible with NSF. There were no differences in the animals treated with zinc, which was introduced to investigate whether transmetallation and subsequent zinc deficiency could be blamed for the clinical and histologic findings. Tissue levels of gadolinium were also investigated by the same technique we used, which was inductive coupled plasma atomic emission spectroscopy. In fact, the Bayer Schering investigators were able to identify the gadolinium within the tissue and quantitate it, just as we had [21].
All of these studies—epidemiologic evidence, clinical description, pathologic evidence, and animal evidence—indicate that there does seem to be a gadolinium association with NSF.
Summary Opinion
In all things it is a good idea to hang a question mark now and then on the things we have taken for granted.
—Bertrand Russell
In my opinion, we can hang a few question marks. These are my opinions:
References
- . Scleromyxoedema-like cutaneous disease in renal-dialysis patients. Lancet. 2000;356:1000–1001
- . Nephrogenic fibrosing dermopathy. Am J Dermatopathol. 2001;23:383–393
- Cowper SE. Nephrogenic Fibrosing Dermopathy [NFD/NSF Website]. 2001-2007. Available at http://www.icnfdr.org. Accessed 11/16/2007
- . Nephrogenic Fibrosing dermopathy: the first six years. Curr Opin Rheumatol. 2003;15:785–790
- . Nephrogenic fibrosing dermopathy (scleromyxedema-like illness of renal disease). J Am Acad Dermatol. 2003;48:55–60
- . Clinical and pathological features of nephrogenic fibrosing dermopathy. J Am Acad Dermatol. 2003;48:42–47
- . Nephrogenic fibrosing dermopathy: suspect identified, motive unclear. Am J Dermatopathol. 2003;25:358
- . Fibrosing skin condition among patients with renal disease—United States and Europe 1997-2002. Morbid Mortal Wkly Rep. 2002;51:25–26
- . Nephrogenic fibrosing dermopathy with systemic involvement. Arch Dermatol. 2003;139:903–906
- . Gadolinium—a specific trigger for the development of nephrogenic fibrosing dermopathy and nephrogenic systemic fibrosis?. Nephrol Dial Transplant. 2006;21:1104–1108
- . Gadolinium—a specific trigger for the development of nephrogenic fibrosing dermopathy and nephrogenic systemic fibrosis?. [erratum] Nephrol Dial Transplant. 2006;21:1745
- . Pharmacokinetics of gadodiamide injection in patients with severe renal insufficiency and patients undergoing hemodialysis or continuous ambulatory peritoneal dialysis. Acad Radiol. 1998;5:491–502
- . Safety of gadolinium contrast agent in hemodialysis patients. Acta Radiol. 2001;42:339–341
- . An investigation of the toxicity of gadolinium based MRI contrast agents using neutron activation analysis. Magn Reson Imaging. 2003;21:541–544
- Nephrogenic systemic fibrosis: suspected causative role of gadodiamide used for contrast-enhanced magnetic resonance imaging. J Am Soc Nephrol. 2006;17:2359–2362
- . Gadolinium and nephrogenic fibrosing dermopathy in pediatric patients. Pediatr Nephrol. 2007;22:1395
- . Nephrogenic systemic fibrosis: a review of 6 cases temporally related to gadodiamide injection (Omniscan). Invest Radiol. 2007;42:139–145
- . European Society of Urogenital Radiology guidelines on contrast media application. Curr Opin Urol. 2007;17:70–76
- . Gadolinium is detectable within the tissue of patients with nephrogenic systemic fibrosis. J Am Acad Dermatol. 2007;56:21–26
- Cowper SE. Nephrogenic systemic fibrosis: a ten year overview. Presented to: US Food and Drug Administration; Silver Spring, Md; February 1, 2007.
- . Gadolinium is quantifiable within the tissue of patients with nephrogenic systemic fibrosis. J Am Acad Dermatol. 2007;56:710–712
PII: S1546-1440(07)00448-6
doi:10.1016/j.jacr.2007.08.013
© 2008 American College of Radiology. Published by Elsevier Inc. All rights reserved.
Volume 5, Issue 1 , Pages 23-28, January 2008
