The 510(k) clears the range, not the fit

Most 510(k) clearances cover one design. Build the cleared device again next year and it is, within manufacturing tolerance, the same shape as the one FDA reviewed. A patient-matched device — a cranial plate, an orthopedic guide, an implant generated from a specific patient’s own imaging — is cleared differently, on purpose, and the public record never says so in those words.

What FDA actually reviews

FDA’s Technical Considerations for Additive Manufactured Medical Devices guidance, finalized December 5, 2017, lets a patient-matched device be produced within a design — or performance — envelope set before any patient-matching starts: a defined list of clinically relevant parameters, each with a minimum and a maximum, and a statement of which ones a specific patient’s file is allowed to move within. A 510(k) built this way does not ask FDA to review one shape. It asks FDA to review a boundary.

Substantial equivalence, shown at the edges

A sponsor doesn’t test every shape the envelope could ever produce — it tests the worst case: the configuration sitting at the envelope’s own edges, the thinnest wall, the sharpest feature radius, the largest surface area the stated range allows. FDA’s own webinar on the guidance works through this with a cranial-plate example, treating substantial equivalence as shown for the envelope once those boundary configurations pass. The actual plate generated for any one patient, built somewhere inside that envelope and almost never sitting exactly on a tested edge, is never itself the subject of a submission. It was cleared by belonging to a boundary FDA already reviewed, not by being reviewed on its own.

A fixed-design device is cleared because FDA looked at it. A patient-matched device is cleared because FDA looked at everything it could ever become — which means no reviewer, ever, looks at the one a specific patient actually gets.

Not the custom device exemption

This is not the same gap the custom device exemption leaves. A device meeting 21 U.S.C. § 360j(b)’s conditions skips 510(k) review entirely — no K-number, nothing in openFDA, by statute. A patient-matched device clears the ordinary way: it gets a K-number, a product code, an entry in openFDA’s 510(k) record, same as any other clearance. The gap sits on the other side of that record. The clearance event is real and public. What it covers is a range, and the record has no field for where inside that range any specific unit actually fell.

The practice

A portfolio or competitor watch built on K-numbers sees a patient-matched clearance exactly once — the day FDA cleared the envelope. It will never see another event for that device type again, no matter how many patient-specific units get built under that one clearance for years afterward, because none of them is a new submission. The record of which patient got which configuration, built and validated against the envelope at build time, lives in the manufacturer’s own design history file under 21 CFR 820.30(j) and device history record under 21 CFR 820.184 — never in any FDA database a monitoring feed, this one included, could read. Per what we monitor and how often, FDA Radar reads the clearance event FDA actually published; whether a given patient’s unit stayed inside the envelope FDA reviewed is a quality-system question for whoever holds that file, never a judgment this makes for them.