FDA proposes PMA alternative for digital mammography systems

Agency claims full-scale study needed for 510(k)

Some full-field digital mammography developers received a shock this month when the Food and Drug Administration’s Center for Devices and Radiological Health articulated its latest position on the digital mammography approval process. In faxed letters to those manufacturers with which it has already been in communication regarding the technology, the CDRH recommended a surprising alternative to 510(k) applications: premarket approvals, or PMAs.

The agency’s suggestion is an attempt to offer manufacturers a new path to clearance, since, the CDRH believes, establishing equivalence between digital and film-screen mammography, as required for a 510(k), is difficult to do without a full-scale clinical trial. The FDA has been sharply criticized in recent years for its digital mammography guidelines, which require vendors to establish equivalence between full-field digital and film images within a 90% to 95% correlation range. Industry and clinical experts have repeatedly emphasized that this correlation is almost impossible, due to the human variability between readings of the same film, and have argued for speedy clearance of what they believe to be a better technology than conventional film-screen.

Once published, the FDA’s new guidance for pursuing digital mammography clearance is sure to cause consternation among digital mammography developers. Industry watchers fear that manufacturers will have to begin again to gather clinical data appropriate for the PMA, and therefore will be further delayed in achieving regulatory clearance—and will incur hefty added expense. Representatives from several vendors developing digital mammography systems were unavailable for comment on the new guidance.

At a meeting last year between the agency and authorities on digital mammography, the FDA heard several clinical study ideas: that companies submit digital and analog images for side-by-side feature analysis; that companies conduct noninferiority studies using 500 to 1000 subjects, with digital and film-screen images taken from all positive patients and a subset of negative patients; and that companies conduct large-scale, 20,000-patient studies to establish the efficacy of digital mammography devices (SCAN 9/16/98).

After assessing these and other clinical models that attempt to establish the equivalence of full-field digital mammography to film-screen, the FDA has come to the conclusion that the full-scale study is crucial to a 510(k) application. But since many vendors are unwilling or unable to undertake this kind of study, the agency plans to suggest the PMA as an option for clearance, according to Dr. Susan Alpert, director of the FDA’s Office of Device Evaluation. A PMA does not require vendors to establish equivalence to existing technology like the 510(k) does, but does require them to prove the efficacy and safety of the product under review.

“We’re not saying that no one can do the 510(k), that no one can come up with data from a trial that would establish equivalency,” Alpert said. “If a company can come up with an effective trial design and justify it, more power to (them). But we’ve concluded that we can’t figure out a way to do a 510(k) without a full-scale study, and we’re offering the PMA as a regulatory option.”

The FDA has been cautious in its approach to digital mammography, not because of the technology itself, but because of the evaluative tools available to measure the technology, and because of mammography’s screening function..

Using the PMA for digital mammography is not a new idea for the FDA. In fact, the agency seriously considered naming it the preferred application process when it was preparing its initial draft guidance for the technology in 1995 (SCAN 9/13/95). But a year later, the FDA published guidance that pointed vendors in the direction of the 510(k) application. The guidance outlined both a more stringent and a more relaxed 510(k) process than expected: Digital mammography developers were required to submit supporting clinical data not required for other 510(k)s, but were not required to conduct large-scale studies.

Even with the published 1996 guidance, the PMA has always been an option for digital mammography developers, Alpert said. Yet vendors tend to avoid it, viewing it as a more time-consuming, rigorous process than the 510(k). Since the PMA covers devices considered higher risk, it carries requirements that the 510(k) process does not: Applicants must provide manufacturing information in more detail than is required for a 510(k), and must make annual reports on any design changes to the equipment under review, perhaps conducting additional clinical studies to evaluate the effect of the changes on the device.

After last year’s feedback meeting, the FDA said that it would publish new guidance that took industry experts’ critiques into consideration. The agency expected to publish that guidance this month, but has again been delayed as it incorporates its new thinking into the document. When published, the guidance will most likely include a recommendation to vendors to consider the PMA, as well as suggestions for various types of clinical trials, including the full-scale study the agency believes is necessary to prove equivalence with film-screen.

But companies shouldn’t be intimidated by the PMA, according to Alpert. Even before the 1997 passage of the FDA Modernization Act (FDAMA), the agency had been taking steps to streamline its regulatory review process. Last October, the CDRH boasted that for the second year in a row, its premarket review program had no backlog of 510(k)s, PMAs, or PMA supplements for fiscal year 1998, which the agency attributed in part to the fact that more companies were consulting with the FDA as they planned clinical studies (SCAN 10/28/98). Although PMA clearances tend to take at least 12 months, Alpert said that companies with mature development programs for digital mammography might receive clearance in less time.

“All the clinical studies (that) companies have been doing are what make the PMA doable,” she said. “Vendors can use the work they’ve done thus far in a PMA application. Their work is not a loss, and it may be sufficient to support a PMA.”