A 48-year-old black man, on hemodialysis since August 2002, presented to his primary care provider (PCP) in July 2006 with excruciating leg pain. According to the patient, the leg pain had worsened during the previous six months and was so severe that he was barely able to walk without pain. He was a full-time night security guard and reported walking three to five miles each night.

The man was undergoing hemodialysis three times per week, necessitated by nephritic range proteinuria. He had a questionable history of diabetes but a known diagnosis of hypertension. Definitive diagnosis through kidney biopsy was not obtained because of the associated risk, the patient's obesity, and his aversion to the procedure.

The patient had recently been hospitalized with shortness of breath and fluid overload. Intensive dialysis allowed a significant drop in his dialysis target weight. He was readmitted a few days later with chills, fever, cough, and shortness of breath. He was diagnosed with bilateral pulmonary emboli. The patient said his hypercoagulation work-up was negative, but he was started on warfarin before discharge.

On current presentation, he had swollen, tender legs and multiple excoriations over the calves, explained by the patient's admitted scratching. His skin was shiny and tight. He was still taking warfarin, with an international normalized ratio of 2.1. The patient denied shortness of breath, pruritus (any more than expected with renal disease), or increased fluid.

In addition to warfarin, he was taking esomeprazole 40 mg/d, extended-release metoprolol 25 mg bid, cinacalcet 90 mg/d, sevelamer 4,000 mg and lanthanum 5,000 mg before every meal, mometasone furoate as needed, hydroxyzine 25 mg every four hours as needed, miconazole powder applied to the feet as needed, and a daily prescription multivitamin complex.

Laboratory tests included normal findings (for a dialysis patient) on the complete blood count; blood urea nitrogen, 101 mg/dL (reference range, 7 to 20 mg/dL); serum creatinine, 16.6 mg/dL (0.8 to 1.4 mg/dL); Kt/V (a measure of adequacy of dialysis), 1.37 (acceptable); calcium, 9.6 mg/dL (8.2 to 10.2 mg/dL); serum phosphorus, 5.6 mg/dL (2.4 to 4.1 mg/dL); intact parathyroid hormone, 359 ng/L (10 to 65 ng/L).

The patient's PCP prescribed oxycodone for the pain and referred him to the vascular clinic for evaluation of his legs. A lower leg duplex scan with ankle/brachial indices performed on July 18 showed significant bilateral peripheral vascular disease. Subsequent magnetic resonance angiography (MRA) showed a questionable adrenal gland mass. Abdominal CT with and without contrast yielded negative results for the adrenal mass but showed a cyst in the right kidney. Although cysts are commonly found in dialysis patients, the vascular surgeon elected to evaluate the cyst with an MRI with gadolinium; the mass was found to be hemorrhagic.

Further vascular work-up continued, including MRI with gadolinium on September 26, 2006, which revealed two-vessel runoff in the right foot and three-vessel runoff in the left foot. According to the vascular consult, there was no area to bypass. The patient was sent back to his PCP. At this point, he was taking oxycodone four times per day and continuing to work full-time as a night security guard.

The patient was then sent to neurology for evaluation. By this time, the severity of his leg pain had increased 90%, with worsening swelling and persistent shininess (see figure). The neurologist was unable to obtain electromyograms due to the severity of the patient's pain and lower extremity swelling. No definitive diagnosis could be made.

About one year later, the man's attending nephrology group received copies of the work-up that the PCP sent to the dialysis center. It was apparent that neither the patient's PCP nor the vascular, radiology, or neurology consultants had seen the FDA warning released in June 2006¹ regarding the use of gadolinium in patients with renal disease. What had started out as a peripheral neuropathy (either renal or diabetic in etiology) was now a full-blown case of nephrogenic systemic fibrosis (NSF).

Open biopsy performed on October 29, 2007, confirmed the presence of gadolinium in the patient's epidermis. He became the first documented case of NSF in the Washington, DC area.

Discussion
In the late 1990s, several reports of an unknown sclerosing dermopathy in patients with chronic kidney disease began to emerge. In 2000, the new entity was named nephrogenic systemic fibrosis, with a disease course demonstrating systemic involvement that affected multiple organ systems and often resulted in severe joint limitations. A Web-based reporting system for this newly described disease, created by Shawn Cowper, MD, of Yale University,² made it possible to investigate associated epidemiologic factors.

Neither gender, race, nor age appeared relevant. However, all patients had renal disease—acute, chronic, or transient—and more than 90% of patients were dialysis dependent. Factors since recognized to confirm a diagnosis of NSF are severe renal impairment (ie, glomerular filtration rate [GFR] < 30 mL/min/1.73 m²),³ CD34+ dendritic cells found on deep biopsy,⁴ and the following clinical manifestations:

• Skin. Burning or itching, reddened or darkened patches; possible skin swelling, hardening, and/or tightening.

• Eyes. Yellow raised spots in the whites of the eyes.

• Bones, joints, muscles. Joint stiffness; limited range of motion in the arms, hands, legs, or feet; pain deep in the hip bone or ribs; and/or muscle weakness.³

Theories abounded on the cause of NSF. While the presence of renal disease is a requirement, dialysis did not seem to be.⁵ Ten percent of NSF cases are patients who have never been dialyzed, and thousands of dialysis patients never develop NSF. Neither was any temporal correlation to dialysis found: While some patients developed NSF soon after starting dialysis, many had been on dialysis for years before NSF occurred. No association was found between NSF and the type of dialysis (inpatient, outpatient, hemodialysis, or peritoneal dialysis), the filter, manufacturer, dialysate, technique, or dialysis unit.²

Authors of a retrospective study involving two large tissue repositories looked for cases of NSF before 1997, but none were found.⁶ If dialysis was not causing NSF, and the disease did not appear to have existed before 1997, what renal toxin had been introduced in the 1990s to explain it?

One early suspicion involved erythropoietin (EPO), used to treat anemia in patients with kidney disease. Skin changes had been reported in some patients after initiation of treatment with EPO, and the NSF patients received a significantly higher mean dose of EPO than controls received.⁷

Ninety percent of patients with NSF had fistula reconstruction or dialysis catheter placement, but these are common in renal disease patients.⁸ Forty-eight percent of patients had had liver or kidney transplants, and 12% had hypercoagulable states. Most patients with NSF had never received ACE inhibitors. Were the protective antifibrogenic properties of these agents missing?

Mystery Solved
In a triumph for the Internet and its capacity to disseminate information around the world, a breakthrough came in 2006 from a small town in Austria. Grobner⁹ described nine patients who had received gadodiamide (Omniscan™)–enhanced MRA, five of whom developed NSF. Upon release of this report, researchers reexamined the original cases and detected a clear correlation between gadolinium and NSF. Because the contrast dose given for MRA can be as much as three times that required for routine MRI, the absence of NSF cases before 1997 suddenly made sense.

In May 2006, researchers for the Danish Medicines Agency reported 13 cases of NSF in patients injected with gadodiamide.¹⁰ Within months, 28 biopsy-proven cases were reported in St. Louis, six in Texas, and 13 at the University of Wisconsin—all involving patients exposed to gadolinium.^11-13 It was apparent that NSF was iatrogenic and could be controlled.

What We Have Learned Since
In subsequent research, it has been found that more than 90% of reported cases of NSF occurred following exposure to gadodiamide—although gadodiamide accounts for only 15% of all gadolinium injections worldwide,¹⁴ and this number is decreasing as more cases are reported. The correlation between gadodiamide and NSF is so strong that its manufacturer, GE Healthcare, sent practitioners a letter in June 2006 warning of NSF as an adverse effect of gadolinium exposure.¹⁵ Two days later, the FDA issued an advisory on gadolinium-enhanced imaging procedures, recommending prompt hemodialysis after gadolinium exposure and reminding radiologists and nephrologists that gadolinium is not FDA approved for MRA.¹

Although the 44% incidence rate of NSF reported by Grobner⁹ has never been replicated, a retrospective review of all known NSF cases affirmed that more than 90% of patients had been exposed to gadolinium.¹⁴ Two 2007 reports published in the Journal of the American Academy of Dermatology demonstrated that gadolinium was detectable in the tissues of patients with NSF.^16,17

In Europe, in response to the May 2006 report from the Danish Medicines Agency,¹⁰ the European Society of Urogenital Radiology revised its guidelines with a directive that gadodiamide not be administered in any patients who had reduced kidney function or were undergoing dialysis.¹⁸ Shortly thereafter, the European Committee for Medicinal Products for Human Use issued a contraindication for gadodiamide use in patients with severe renal impairment and advised that these patients not be given gadolinium unless there was no other choice.¹⁹ A contraindication was also issued for gadodiamide use in patients with previous or anticipated liver transplantation.

The American College of Radiology guidelines published in 2007²⁰ stated that patients with any level of renal disease should not receive gadodiamide.

In March 2007, GE Healthcare published a paper on NSF, reiterating the safety of gadodiamide while acknowledging that 120 more cases had been reported to them ("usually associated with exposure at high doses").²¹ The FDA upholds an alert regarding use of all gadolinium-based contrast agents for patients with acute or chronic severe renal insufficiency,³ while stopping short of a ban on gadodiamide in such patients.

How Common Is NSF?
In a 2007 study conducted at the University of Wisconsin, Sadowski et al¹³ reported 13 cases of gadolinium-induced NSF, 11 involving patients with a GFR below 30 mL/min/1.73 m² but two with a GFR between 30 and 60 mL/min/1.73 m² (ie, with renal insufficiency, although the authors noted that renal insufficiency was acute in these two patients). The incidence of NSF was 4.6% among hospitalized patients with a GFR be-low 60 mL/min/1.73 m² who underwent gadolinium-enhanced MRI at the university hospital's radiology department. A reexamination of the charts of the patients with a GFR between 30 and 60 mL/min/1.73 m² revealed that these patients had levels below 30 mL/min/1.73 m² when their gadolinium exposure took place.

In an outpatient population–based calculation performed by Deo et al,²² a 2.4% chance of NSF was determined for each gadolinium exposure. Incidence of NSF was calculated at 4.3 cases per 1,000 patient-years in this population, making NSF as common as contrast-induced nephropathy. Nearly 5% of patients with NSF have an exceedingly rapid and fulminant disease course that may result in death. NSF, of itself, is not a cause of death but may contribute to death by restricting effective ventilation or by restricting mobility to the point of causing an accidental fall that may be further exacerbated by fractures and clotting complications. NSF survivors may experience disabling systemic symptoms. Full recovery occurs only in patients who recover renal function, either naturally or by kidney transplantation.⁴

Why Is NSF More Common With Gadodiamide?
As of June 2008, five gadolinium-based contrast agents were FDA approved for use with MRI (none with MRA)³: gadobenate (MultiHance^®), gadodiamide (Omniscan), gadopentetate (Magnevist^®), gadoteridol (ProHance^®), and gadoversetamide (Opti-MARK^®). More than 90% of NSF cases are associated with gadodiamide. Because this agent is the least stable thermodynamically, it may be more likely than the others to transmetallate.¹⁴ All gadolinium chelates are excreted by the kidney, and the decreased renal clearances associated with renal impairment may expose patients to prolonged gadolinium transmetallation, allowing the agent to accumulate in bone and other tissue.

Gadoterate (Dotarem^®), a cyclic gadolinium-based agent that is available in Europe but not the US, is considered more stable than other agents. It has been suggested that such agents may be safer choices for patients with decreased renal function.^14,19

Strategies to Prevent NSF
In the US and Europe, only a physician who has consulted with a radiologist can write an order for gadolinium use in a patient with a GFR below 30 mL/min/1.73 m².^18,20 European guidelines do not allow use of gadodiamide in such patients.

Although the actual population-based occurrence of NSF is low, the nature of the disease calls for an effort to limit vulnerable patients' exposure to gadolinium (see box). Outside of withholding imaging procedures, the only currently known strategies to reduce the incidence of NSF are to use a more stable, nonchelating gadolinium¹⁴ and to remove the gadolinium as soon as possible.^3,24

It has been recommended that patients with renal disease who are presently undergoing dialysis be dialyzed within two to three hours of gadolinium exposure, then again within 24 and 48 hours, provided it is clinically safe.^20,24 This has been shown to remove 99% of the gadolinium.²³

Since peritoneal dialysis clears gadolinium poorly, hemodialysis is recommended for peritoneal dialysis patients after gadolinium exposure, following the regimen outlined above.²⁰

No consensus has been reached regarding the patient with a GFR between 30 and 60 mL/min/1.73 m², nor for the patient with a lower GFR and no access for dialysis to be administered. Placement of a catheter for two days' dialysis incurs both surgical and renal risks for these patients.⁸

Patient Outcome
The only known cure for NSF is kidney transplantation, which is associated with a complete cure rate of 40%.^4,25 Nevertheless, while this manuscript was in preparation, the patient presented in this case study underwent kidney transplantation. On day 8 postsurgery, he was no longer taking oxycodone, his skin condition was clearing up, and he was feeling considerably better. His health care providers hope for further regression from his disease.

Conclusion
NSF is just one example of iatrogenic conditions that can occur in any hospital, office, or clinic. Health care providers cannot be too vigilant in keeping abreast of warnings from the FDA and other agencies. In this case, several clinicians overlooked a recent, urgent public health advisory, with significant consequences.

Man, 48, With Excruciating Leg Pain