FDA to revamp 510(k) guidelines for digital mammography systems

New rules could affect regulatory submissions of full-field systems

The Food and Drug Administration plans to revise its 1996 guidelines governing 510(k) submissions for full-field digital mammography systems, the agency has said. Digital mammography developers don’t believe the revised guidelines will delay the commercial introduction of full-field systems, and some say they plan to submit 510(k) submissions using the agency’s 1996 criteria.

In a meeting that generated new ideas but no concrete resolution, the FDA’s Radiological Devices Panel met with mammography experts on Aug. 18 to evaluate the existing standards for digital mammography submissions. The meeting reflects the FDA’s ongoing struggle with how to regulate full-field digital mammography systems that are under development by several vendors. Due to the novelty of the systems and the special requirements of mammography, the FDA is reluctant to allow them through the same 510(k) process that most imaging devices follow. At the same time, the FDA has expressed its intent to help vendors get full-field systems to market as quickly as possible.

The FDA’s current guidance was established in 1996 and requires companies to submit a 510(k) application with supporting clinical data, a criterion not required for standard 510(k) submissions (SCAN 9/27/95). The guidelines also require vendors to prove the equivalency of full-field digital images to film within a 90% to 95% correlation range. This standard is unrealistic because of the human variability between readings of the same film, according to Dr. Edward Sickles, chief of breast imaging at the University of California, San Francisco’s department of radiology.

“The problem is that there is more inter- and intra-observer variability in interpreting film mammograms than is allowed for in the guidance document for digital mammograms,” Sickles said. “It’s essentially impossible for a manufacturer to comply with the current guidance.”

As it considers issuing new guidelines, the FDA is weighing a number of clinical study ideas suggested at last month’s meeting. Some participants suggested that companies submit digital and analog images for side-by-side feature analysis by a panel of experts. Another possibility is for companies to conduct noninferiority studies using 500 to 1000 subjects, in which both digital and screen-film images are taken from all positive patients and a subset of negative patients over an interval of time and analyzed statistically. The third option is to conduct large-scale, 20,000-patient studies to determine the full and exact efficacy of digital devices. Both the FDA and meeting participants, however, acknowledged that this last option is burdensome to vendors and should be conducted by an outside agency once units are placed in the market.

The FDA expects the new document to outline various options for clinical trials, rather than calling for one particular kind of trial, according to Bob Doyle, executive secretary of the Radiological Devices Panel and a scientific reviewer in the FDA’s radiology branch. Companies could combine studies to craft the strongest argument in favor of their technology. The agency did not give a timeline for the new guidelines, saying only that they would have to go through the traditional process of public comment and internal review before being released. A typical process of guidance revision such as this takes between 90 and 120 days.

What next for vendors? In the meantime, companies have the option of waiting for the new guidelines to be issued before amassing clinical data to support their submissions, or they can proceed under the 1996 guidelines. Digital mammography developers should not necessarily have to start over with clinical data once the new guidelines are made public, according to Doyle.

“The agency will accept any adequately supported method of demonstrating equivalence, whether it’s in our new guidance or not. If we put out 15 different ways (to prove equivalence), and you come up with number 16, if it’s viable and well-supported, we will accept that,” Doyle said.

At least one full-field digital mammography developer, Fischer Imaging, expects it will move forward with its existing study data compiled under the 1996 agreement guidelines. The Denver-based vendor will analyze its data with the understanding that they reflect inter- and intra-observer variability.

“We have been working on the assumption that we will submit a 510(k) application under the agreement guidelines,” said Mike Tesic, vice president of engineering. “We (will) have to take into account the inter- and intra-observer variability and do our statistical analysis with those factors included.”

Only one company, Trex Medical of Danbury, CT, has already submitted a 510(k) to the FDA, along with clinical data from a study that included 520 women. Trex has not received clearance for its system yet, but company executives do not believe the FDA’s latest discussion will influence its progress.

“(The new guidances) will probably not affect us,” said Richard Bird, director of product clinical development at Trex. “(The FDA is) still reviewing our 510(k) independently of what they’re doing with the guidance, in my understanding.”

The clearance process has taken so long because of the nature of mammography, according to Doyle. Other forms of diagnostic radiology incorporate additional information about the patient’s condition, including results from other modalities. But mammography is the only imaging method used to diagnose breast cancer, and because it is a screening tool, the patients are asymptomatic.

“You’re making a decision based on the image alone, whether that patient should have a further workup, or go home and come back a year later. That could be a critical decision, because if you missed a starting cancer and sent the patient home for a year, that could have dire consequences,” Doyle said. “That’s why we’re being much more critical and holding a higher standard for mammography.”