Comments on FDA’s Proposed Rule Governing the De Novo Classification Process

On December 4, 2018, FDA issued a proposed rule that would govern the de novo classification process.  After a comment period, it may be re‑issued as a final rule to take affect 90 days after publication.   While we agree with FDA’s goal of creating greater consistency and predictability in the de novo process, the proposed rule appears in some respects to unduly increase the burden on applicants.  In this regard, the proposed rule would make the de novo process look more like review of a premarket approval (PMA) application than a 510(k) submission.  It is also not clear if FDA has statutory authority to implement certain features of the proposed rule.

As a reminder, de novo classification allows the placement of novel device types into Class I and Class II, rather than burdening them with the Class III, PMA approval process that applies by statutory default to all new device types.  After a new device type is classified in Class I or Class II via de novo classification, similar devices within the same generic type can proceed to market based upon 510(k) clearance.

Most of FDA’s classification regulations were adopted for devices on the market in the 1970s and 1980s.  As time has passed, more novel device types have been developed.  The de novo process is an important way to expand the classification regulations to incorporate new Class I and Class II device types.

Until recently, there was little written guidance from FDA as to how de novo reviews would be conducted.  In recent years, FDA has issued several guidance documents on various aspects of the de novo program (for example: here, here and here).  A regulation would harden “recommendations” in a guidance document into “requirements” that must be followed.  A regulation gives both industry and FDA less flexibility, but it also gives greater certainty about the rules of the road.

In general, the proposed rule is consistent with FDA’s current administrative practice in processing de novo submissions.  The proposed rule appears structurally similar to the 510(k) regulation (21 C.F.R. Part 807, Subpart E) and the PMA regulation (21 C.F.R. Part 814), in setting forth requirements for (i) the format and content of a de novo submission, (ii) the procedures governing FDA’s review, and (iii) grounds for denial.

The proposed content requirements are nearly the same as those FDA proposed in its draft guidance, Acceptance Review for De Novo Classification Requests (Nov. 2017) (see our blog post here).  But, there are several additions that will likely increase the burden on applicants.  For example, de novo sponsors will be required to submit:

• a bibliography of all published and unpublished reports on the device and any other information relevant to a device’s safety or effectiveness;
• samples of the device and its components, if requested by FDA; and
• advertisements for the device.

These requirements are elements of a PMA application, not the 510(k) submission.  21 C.F.R. § 814.20(b)(8)‑(10).  With regard to advertisements, in particular, the PMA regulation is limited to “advertising that constitutes labeling under section 201(m) of the act.”  Id., § 814.20(b)(10).  In contrast, this proposal is broader.  It is not clear what the justification is, since FDA does not have authority to over the advertising of devices classified into Class I or II.  Perhaps they would justify this request based on the need to determine the intended use.  See id., § 801.4 (defining intended use to include labeling and advertising).

Unlike the 510(k) regulations, and more like the PMA regulations, the proposed de novo rule specifies the types of information required in a submission, including various types of non‑clinical data.  Compare 21 C.F.R. § 807.92(b)(1) with id. § 814.20(b)(6)(i).  This specification will at least provide greater certainty about what is required.  One potential omission is acknowledgement of the types of tests done for in vitro diagnostic devices, which have been a significant percentage of de novo clearances.

The procedure for review of submissions in the proposed rule generally follows FDA’s current administrative approach.  Interestingly, the proposed regulation states that a de novo request can be refused for filing if “the requester has not responded to, or has failed to provide a rationale for not responding to, deficiencies identified by FDA in previous submissions for the same device.”  In the preamble to the proposed rule, FDA states that a de novo can be refused for filing if a requester has not provided “a complete response” to such deficiencies.

The completeness of a response is, of course, in the eye of the beholder.  We hope that FDA will clarify in the final rule that a submission will be accepted if there is a response to prior deficiencies, with evaluation of the completeness and adequacy of the response occurring during the review on the merits.  This issue is of particular concern in light of problems in the 510(k) Refuse‑to‑Accept (RTA) process in the early days of its implementation, in which substantive questions were improperly raised.

The proposal sets forth 11 grounds on which the Agency may deny a de novo request.  In the PMA context, the grounds for denial are set forth in the Federal Food, Drug, and Cosmetic Act (FDCA) and are mirrored in the implementing regulation.  In the 510(k) context, the statute defines “substantial equivalence” and the implementing regulation provides that failure to establish substantial equivalence in accordance with this definition is grounds for denial.  In the de novo context, the statute authorizes only two potential grounds for denial:  (i) existence of a predicate device that provides a reasonable basis for a substantial equivalence review; or (ii) a determination by FDA that “the device submitted is not of low‑moderate risk or that general controls would be inadequate to control the risks and special controls to mitigate the risks cannot be developed.”  It seems potentially unauthorized for the implementing regulation to exceed the FDCA by specifying, as it does, nine additional grounds for denial.

Currently, FDA promptly posts letters granting de novo authorization in its de novo database.  Unfortunately, FDA takes considerably longer to post decision summaries, and it has taken months (and sometimes years) before new classification regulations are published in the Federal Register.  For example, the classification regulation for 23andMe’s DEN140044 was published nearly three years after the de novo classification was granted.

More recently, FDA has been posting decision summaries more promptly in the de novo database.  FDA should add a provision to the proposed rule establishing a definite timeline for posting both decision summaries and issuing Federal Register notices.  FDA publishes 510(k) Summaries (and decision summaries for 510(k)s for IVDs) within 30 days of clearance.  That would be a reasonable timeline for publishing the results of de novo classification decisions.

We found it interesting that the proposed rule states that FDA will decide a de novo request within 120 days.  That deadline is already in the statute, so FDA had to adopt the same timeline in the proposed rule.  In actuality, though, FDA routinely misses this deadline.  In the MDUFA IV Commitment Letter, FDA publicly agreed only to approve/deny 50% of de novo requests within 150 days (i.e., 30 days beyond the statutory deadline), with the other 50% presumably being decided somewhere north of 150 days.  In conjunction with the proposed rule, FDA should request appropriate funding from Congress to staff the de novo program to meet the 120‑day statutory requirement.  All parties would benefit.

Possibly the most controversial feature of the proposed rule is that it would give FDA authority to inspect submitter’s manufacturing facilities and clinical trial sites.  Here is the proposed regulatory provision (§ 860.256(c)):

(c) Prior to granting or declining a De Novo request, FDA may inspect relevant facilities to help determine:

(1) That clinical or nonclinical data were collected in a manner that ensures that the data accurately represents the benefits and risks of the device; or

(2) That implementation of Quality System Regulation (part 820 of this chapter) requirements, in addition to other general controls and any specified special controls, provide adequate assurance that critical and/or novel manufacturing processes produce devices that meet specifications necessary to ensure reasonable assurance of safety and effectiveness.

Start with manufacturing inspections:  In the 510(k) context, the FDCA expressly forbids FDA from conducting inspections for Quality System Regulation (QSR) compliance, unless “there is a substantial likelihood that the failure to comply with such regulations will potentially present a serious risk to human health.”  FDCA § 513(f)(5).  FDA has declared that it may conduct preclearance inspections for a few device types, e.g., infusion pumps.  Most device types do not meet this standard and preclearance manufacturing inspections are rare.

In the PMA context, the statute permits FDA to withhold approval if manufacturing facilities do not conform to QSR requirements.  FDCA § 515(d)(2)(C)).  The implementing regulation expressly authorizes conditioning approval on a successful manufacturing inspection.  21 C.F.R. § 814.44(e)(1)(iii).  As a matter of administrative practice, FDA routinely performs QSR inspections prior to granting PMA approval.

As to de novo classification, the FDCA is silent on whether preclearance inspections are authorized.  Yet, in the proposed rule, FDA grants itself authority to conduct manufacturing inspections when deciding de novo classification requests.  Perhaps recognizing this legal weakness, FDA’s proposal is not a straight‑up right to inspect for QSR compliance.  Rather, FDA purports to authorize itself to inspect whether the applicant’s “implementation” of the QSR “in addition to other . . . controls” (whatever that means) will “provide adequate assurance that critical and/or novel manufacturing processes” will “produce devices that meet specifications necessary to ensure reasonable safety and effectiveness” (proposed § 860.256(c)).

The actual meaning of this convoluted language is anybody’s guess.  It seems like a questionable effort to tie QSR compliance to device classification.  There is no support in the FDCA for doing so.  On the contrary, the statutory provision that authorizes all the various classification proceedings (FDCA § 513) does not authorize manufacturing inspections, with an express limited exception in the 510(k) context (discussed above).  In practice, none of the classification regulations promulgated in the 1970s and 1980s were associated with manufacturing inspections.  Since a de novo review is in fact the promulgation of a new classification regulation, it would seem that likewise a manufacturing inspection is not authorized for de novo review.  Certainly, there is no express statement anywhere in § 513 that FDA may conduct manufacturing inspections in conjunction with de novo classification proceedings (in contrast to the express authorization in § 515 in connection with PMA approval).

Apart from the uncertain statutory grounds, this process of inspecting de novo applicants for QSR compliance would create an undue burden on a first comer.  After a de novo is granted, subsequent applicants will proceed through the 510(k) process (for non-exempt devices).  As noted above, FDA is expressly prohibited from inspecting 510(k) applicants for QSR compliance, unless there is a serious risk to health.  Therefore, these second comers will have a lower bar to clearance than the de novo applicant, creating an unevenly applied regulatory scheme.

The proposed rule is on more solid ground with clinical study site inspections.  FDA has authority to inspect data and information related to investigational devices, including the results of clinical studies evaluating such devices.  E.g., 21 C.F.R. § 812.145.  The statutory authority underlying that regulation is not open to question.

One wonders also how practical is will be for FDA to routinely conduct clinical and/or manufacturing inspections in a 120‑day time frame for de novo review?  It would be resource‑intensive to complete inspections in this time frame, possibly detracting from the ability to conduct routine inspections or causing FDA to frequently miss the deadline for completing de novo reviews.  With regard to manufacturing, there will also be a substantial burden on companies that have developed novel device types.  These firms are frequently start‑ups that have not completed building out their manufacturing facilities and procedures.  It might be better to conduct an early inspection once they are engaged in actual commercial distribution.

As to clinical data, although we do not question FDA’s authority or even the need in some cases to conduct inspections, we are already aware of two past de novo requests for which FDA has inspected clinical data and the review time significantly exceeded the 120-day statutory decision deadline.  It is not clear how routine clinical inspections would work as a practical matter or whether FDA envisions such inspections taking place only if there are “for cause” concerns about the integrity or validity of the clinical data.  This point at least needs to be clarified.

Overall, in our view, the proposed rule is a good idea to bring greater certainty to the de novo process.  If the proposed rule is finalized as‑is, though, it will materially increase the burden on applicants seeking de novo marketing authorization for a low‑risk novel device.  That potential impact is concerning, especially in light of FDA’s recent statements about an intent to increase utilization of the de novo pathway.

Those in industry who would be subject to the proposed rule may wish to submit comments on the legalities and burdens associated with the proposed rule.  Comments are due by March 7, 2019.