Dec. 2, 2009: This post has been corrected.

The U.S. Food and Drug Administration is weighing further regulation of three drugs used to create high-contrast images on MRI scans, based on a new analysis that suggests they carry a higher risk of causing a rare but potentially fatal disease.

The issue -- highlighted in an October story by BusinessWeek and ProPublica -- marks a setback for GE Healthcare (GE), which contends its product is no riskier than competing imaging drugs. FDA reviewers said GE's drug, Omniscan, had a disproportionally high number of reports of the disease compared with its peers.

An FDA advisory panel is scheduled to assess on Dec. 8 whether new restrictions or warnings are in order for the three drugs. The others are Bayer HealthCare’s Magnevist (PDF), the market leader, and Optimark, made by Mallinckrodt.

All three have been associated with nephrogenic systemic fibrosis, or NSF, which can occur in patients with impaired kidney function. The drugs, which contain the metal gadolinium, are injected during magnetic resonance scans. Although the cause of NSF is uncertain, researchers theorize that an inability to eliminate gadolinium may be a factor. About 8 million vials of gadolinium-based imaging agents are sold each year. So far, the number of NSF cases is in the hundreds.

In patients who do contract the disease, however, outcomes can be painful, disfiguring and deadly. The skin thickens and hardens around joints, gradually limiting movement and crippling or incapacitating some victims. The disease also can affect internal organs and set the stage for a lung embolism or other potentially life-threatening condition. The disease has no known cure, and the FDA said the two-year mortality risk for NSF patients is more than double that of kidney patients without the disease.

More than 500 lawsuits over NSF have been filed in U.S. courts against makers of gadolinium-based MRI drugs. The majority target GE and Omniscan, and the drug has been the one most cited in NSF reports filed with the FDA and European regulators.

Many major hospitals across the country stopped using Omniscan once the link to NSF first appeared, and in the last two years reports of new NSF cases have virtually stopped as warnings by regulators, manufacturers and doctors took hold.

The new FDA assessment suggests a possible shift in agency policy.

After a link first emerged in early 2006 between NSF and scanning drugs, the FDA treated all the agents -- seven are now sold in the U.S. -- as a class. That put the agency at odds with European regulators, influential American radiologists, and two of its own medical reviewers, all of whom had flagged Omniscan as a riskier agent.

GE has relied on the FDA's stance to reinforce its scientific and marketing messages about Omniscan's safety. In court and at the FDA, the company has argued against singling out its drug from competing gadolinium-based agents. "There is no definitive evidence establishing that the risk of acquiring NSF is greater for one agent than for the others in the class," the company says in a new filing to the advisory panel.

Although GE has not proposed a labeling change, the company advises against using Omniscan in scans of the sickest kidney patients.

The FDA has asked the panel for comment on its latest risk assessment and on filings submitted by the manufacturers. In the past, the agency has tended to follow recommendations from the panel, comprising outside doctors and scientists.

Bayer, in its filing, said the NSF risk for Magnevist is lower than that of Omniscan and Optimark. Mallinckrodt, Optimark's manufacturer, told the FDA it was revising package labeling to contraindicate, or recommend against, use of Optimark in patients with the most serious kidney impairment. European regulators already have contraindicated use of Omniscan, Magnevist and Optimark for patients in that class.

In 2007, two FDA medical reviewers independently recommended that Omniscan be contraindicated, as first disclosed in the October story by BusinessWeek and ProPublica. But their boss, R. Dwaine Rieves, rejected the findings. He said his 2007 decision to treat all the drugs as equally risky was based in significant part on "a very limited database."

Rieves, who heads the FDA division that oversees imaging drugs, again will help to decide whether the agency should take action based on the panel's input and the agency's new assessment, which relied on a deeper statistical study and additional marketing data.

The review ranks Omniscan highest, by far, on a ratio indicating whether a drug has a "signal for NSF." It also rates Omniscan as the least chemically stable agent, just ahead of Optimark, based on a formula "consistent with real-life conditions." European regulators reported a similar finding on chemical stability in a study two weeks ago.

GE says in its latest FDA filing (PDF) that theories relying on chemical stability "as explanations for the development of NSF are incomplete and flawed." The company also said the relatively high number of NSF cases linked to Omniscan might reflect reporting bias.

The FDA cited 382 cases in which Omniscan was named as the single imaging agent in an NSF report. Magnevist was named in 195 reports and Optimark in 35.

The FDA put Omniscan's market share from 2005 through 2007 at only 32 percent, compared with 49 percent for Magnevist and 10 percent for Optimark. In the first half of this year, Omniscan's market share had fallen to 17 percent, the agency said.

Scientists say NSF remains a mystery whose solution could open new insights into related diseases that affect many Americans, such as heart disease. But research into the NSF has been slow to gain steam, and the FDA's review singled out GE for tardiness.

The company is the only drug maker that has yet to comply with a 2½-year-old agency mandate that each manufacturer conduct trials on 600 patients with severe kidney disease and 400 with moderate impairment, a crucial step to solving the mystery of NSF.

"Omniscan has not presented a final protocol for studying" the most severe patients, while all the other drugs are at more advanced stages, the FDA said.

The potential risks of such testing may explain why.

In healthy patients, the kidneys quickly excrete gadolinium agents. So far, the universe of NSF patients is almost entirely patients who had severe kidney disease, which affects a few hundred thousand Americans. A handful of NSF cases are patients with more moderate kidney disease, a condition affecting several million Americans.

Doctors involved with NSF say the FDA's requirement to test patients with severe kidney disease could be problematic given the known association with Omniscan.

"It would be unethical for any institutional review board or any physician to approve any such study today for Omniscan with that patient population," said Dr. Emanuel Kanal, a University of Pittsburgh professor of radiology who heads the American College of Radiology's blue ribbon panel on magnetic resonance safety.

GE said that in contacting 684 research centers, none would agree to sponsor a test for a patient with severe kidney disease; only four agreed to trials with moderately impaired patients. With the FDA's approval, the company is looking to conduct the latter trials in Europe and Asia.

Correction (Dec. 2, 2009): An earlier version of this story incorrectly reported that GE's Omniscan had 58% of the U.S. market for gadolinium-based contrast agents between 2002 and 2007. The company says the 58% figure, which it reported to the U.S. Food and Drug Administration, actually represents the share of Omniscan's worldwide sales inside the U.S.