510(k) or De Novo? Choosing the FDA Route for a UK-Built Surgical Robot

To sell a surgical robot in the United States, a UK manufacturer usually needs either a 510(k), built on substantial equivalence to a US predicate and filed at least 90 days before commercial distribution, or a De Novo classification, used when no predicate exists. With a De Novo, FDA must issue its decision within 120 days of receipt. This article draws only on the FDA texts indexed in our FDA surgical robotics and SaMD knowledge base. That base does not cover the MHRA regime, UKCA or CE marking, so nothing here should be read as a statement about UK or EU requirements.

Start from the US market, not your UK or EU file

An FDA strategy rests on US concepts that have no direct equivalent in the documents we hold for other jurisdictions. The central one is the predicate: under 21 CFR 807.92(a)(3), as quoted in FDA's 510(k) Program guidance (July 28, 2014), it is a device legally marketed in the US before 28 May 1976, a device reclassified from class III to class II or I, or a device already found substantially equivalent through a 510(k).

If there is no such device, the FD&C Act places a new device automatically in class III (section 513(f)(1)), whatever its real level of risk. The De Novo guidance (issued 5 October 2021) explains that Congress created the De Novo process so that devices whose risks can be handled by general or special controls are not forced through premarket approval (PMA).

The 510(k): proving substantial equivalence

A 510(k) shows that the new device is substantially equivalent to a predicate. Under section 513(i) of the FD&C Act, it must have the same intended use and either the same technological characteristics, or different ones that do not raise different questions of safety and effectiveness and are shown to be as safe and effective. The guidance stresses that the review is comparative, unlike a PMA's independent demonstration of safety and effectiveness.

FDA's draft guidance on Robotically-Assisted Surgical Devices (RASDs), issued on 25 September 2026 and marked “Draft: Not for Implementation”, adds robot-specific expectations for 510(k)s. They include identifying the classification regulation and product code on the basis of the predicate, giving side-by-side comparison tables, and reserving the words “identical” or “same” for features that really are identical. Because it is still a draft, treat it as an indication of FDA's direction, not as settled policy.

A word of caution on classification. The base includes 21 CFR 876.1500 (endoscope and accessories). Its text is generic and never uses the words “robotic” or “computer-controlled”, so it cannot on its own confirm how a given robot is classified. The draft RASD guidance points to FDA's Q-Submission Programme when you are unsure whether your device is a RASD at all.

The De Novo: when there is nothing to compare against

De Novo leads to class I or class II for devices that have no legally marketed predicate but whose safety and effectiveness can be reasonably assured by general controls, or by general and special controls. You may file it directly (a “Direct De Novo”) or after an NSE (not substantially equivalent) decision on a 510(k).

A granted De Novo lets you market straight away and turns your device into a predicate for future 510(k)s, competitors' included. A declined request leaves the device in class III: the guidance's options are then a PMA or a new De Novo request with more information. The base covers PMA only indirectly. It holds no PMA guidance, so it cannot give you PMA timelines or fees.

Clinical evidence gathered in the UK: what the draft RASD guidance says

This is often the decisive point for a British company. The draft RASD guidance notes that clinical data are often needed for new RASDs, meaning robots that have never been authorised in the US. For investigations conducted outside the US and used to support a US submission, it points to the requirements of 21 CFR 812.28. It adds that, to support US marketing, the data should be representative of the intended US patient population and of how the robot will be used in a typical US operating room.

• Show that non-US patient cohorts have similar demographic characteristics and comorbidities to the intended US population.
• Show that non-US investigators and surgeons have similar training, experience and practice patterns to the targeted US user groups.
• Evaluate how different user groups (surgeons, bedside assistants, scrub nurses) interact with the system and whether the training works.
• Plan for the IDE framework if you study the device in the US: FDA generally considers investigational RASDs to be significant risk devices.

The draft also lists nine surgical endpoints FDA generally recommends, including length of hospital stay, conversion rate, and adverse events, readmissions and reoperations through 30 days. Building them into a UK study from the outset is far easier than reconstructing them afterwards.

If your robot includes AI features you expect to retrain after launch, factor that into the route as well. FDA's final guidance on Predetermined Change Control Plans says a PCCP can be established through a 510(k), a De Novo request or a PMA, so the planned changes are authorised together with the device rather than in a later submission. The draft RASD guidance itself points to a PCCP where post-authorisation software changes are expected.

A decision checklist for UK teams

1. Search FDA's public databases for a US predicate and keep a record of your search terms. The De Novo guidance expects them in a Pre-Submission.
2. Fix your US intended use and indications. A different intended use can by itself make a device NSE.
3. List the technological differences from the closest candidate predicate and judge whether each raises different questions of safety and effectiveness.
4. If no predicate fits, prepare the De Novo risk analysis: each risk to health and its mitigation, labelled as a general or special control.
5. Check that you are within the draft RASD scope. Pre-operative planning and stereotaxic systems are excluded, as are remotely teleoperated and autonomous robots.
6. Book a Pre-Submission. FDA “strongly recommends” one before a De Novo, especially for devices it has not reviewed under a 510(k).
7. Remember the eCopy: a De Novo request without a valid electronic copy is placed on hold and the clock does not start.

To test your own situation against the sources, ask the base questions such as “If there's no legally marketed predicate but special controls can manage the risks, which pathway applies, and how long does FDA have to decide?” Each answer quotes the passage it relies on.

Primary sources: FDA 510(k) Program guidance, FDA De Novo guidance and the draft RASD guidance. Nothing here is regulatory advice for a particular device.