3D mammography has sat in the FDA's highest risk class for fifteen years, and not because anybody decided it belonged there

Digital breast tomosynthesis has been in the Food and Drug Administration's highest risk class since before the first one existed, and no regulator ever looked at the technology and decided it belonged there.

That is how the statute works. Under section 513(f)(1) of the Federal Food, Drug, and Cosmetic Act, a device that was not on the market before 28 May 1976, has never been reclassified, and has never been found substantially equivalent to something that was, lands in class III automatically. Class III means premarket approval, the heaviest submission the agency has. On Friday morning FDA filed a proposed order to move the whole device type out of it.

The record it is built on

Fifteen years of it, and the agency lists the pieces.

The first system, Hologic's Selenia Dimensions 3D, was filed as an application in January 2008, went before the Radiological Devices Panel in September 2010, and was approved on 11 February 2011. Three more original approvals followed, along with 26 supplements. One of those was a panel-track supplement in 2012 that added synthesised two dimensional images, so that a screening exam could produce both views from a single acquisition rather than exposing the patient twice.

That supplement drew one dissenting panel member, who was uneasy that the clinical study had excluded patients with large breasts, implants or tissue markers. FDA says the intervening data has answered the point.

The safety record it cites is thin in the way an agency likes. A search of the recalls database on 21 July returned five Class II recalls and no Class I or Class III recalls. The causes were a reconstruction error on breasts thicker than 90 mm that nearly covered the detector, a risk of the c-arm striking somebody, a software fault affecting image quality in certain modes, an unapproved slabbing function left enabled, and internal bolts working loose over time. No injuries were reported from any of them.

What replaces premarket approval

Not nothing, which is the part worth reading carefully.

Class II means special controls, and FDA proposes to codify them in a new regulation at 21 CFR 892.1717 alongside a list of ten probable risks to health: corrupted images, misinterpretation, inadequate breast coverage, inappropriate compression, device failure, use error, excessive x-ray exposure, interference with other devices, adverse tissue reaction and infection. The controls run to bench testing of imaging characteristics, task-based assessment of diagnostic accuracy using human subjects or phantoms or in silico methods, software verification and validation, clinical image evaluation and specific labelling.

FDA also says, in terms, that it is not proposing to exempt the device type from premarket notification. A new system would still need a 510(k) and a clearance before it could be sold.

Status

A proposed order is not an order. Comments run 60 days from publication and go to docket FDA-2026-N-7630, and the agency proposes that any final order take effect 30 days after it publishes.

The Mammography Quality Standards Act, which governs the facility and the physician rather than the machine, is not touched by any of this.