Monday, August 4, 2014


On July 31, 2014, the FDA released an armful of documents announcing its extensive plans to regulate laboratory developed tests (LDTs).

Update: The FDA's draft guidance was officially released for 120 days of public comment on October 3, 2014 (here).  The official comment period documents are virtually unchanged from the July early release versions.

In this post, we review the major documents released and briefly summarize the back story, which dates to the 1990s.   We discuss some lower-visibility aspects of the documents that may have significant strategic implications.   We note that while the FDA has played its cards with a big splash, the proposal is massive and would be colossal to implement in its full fledged form – as well as raising substantially higher regulatory barriers to daily medical lab operations than in other advanced countries such as Canada, England, and the European community.   The plans might be backed down to something more akin to the New York State special test-by-test approvals that many labs already navigate, but only time will tell.

[Updates through August 6]
[For my summary of September 9, 2014 Hill hearings on FDA-LDTs, see my later blog, here.]


The FDA releases most new guidances spontaneously, but Congress required it to provide 60-days notice to Congress before releasing any LDT regulatory guidance.  Accordingly, the FDA released two lengthy guidance documents on its regulation of LDTs.   The first is a broad 41-page general framework for LDT regulation, and the second gives 25 pages of details on how the FDA proposes its registration, listing, and adverse event reporting features would work.   

Our firm's official and more comprehensive review 
of the FDA documents is here at  

In addition, the FDA released its responses to three pro-and-con LDT Citizen’s Petitions filed in 2006, 2008, and 2013, respectively.  I will begin by discussing these, since they are foundational to the validity of the higher-profile LDT guidance documents.   

If you strongly support the FDA's effort, we note that the FDA will face a range of "gray areas" in actually shifting from the simple world of paper proposals to real-world implementation.  If you don't like the FDA's effort, the gray areas may sound like just the tip of the iceberg to you.    


There is an arcane history dating back to the 1960s and even earlier that “drugs” were oddly defined by Congress in a way that appeared to include “devices.”   (See a 1969 Supreme Court case, here.)   Despite this legal rabbithole, most FDA regulation of devices begins in 1976.

1992: With LDTs, the history can begin with a 1992 Citizen’s Petition authored by noted FDA attorney Jeff Gibbs at Hyman Phelps (here).   Gibbs and his firm argued that the FDA did not have authority to regulate LDTs, and the FDA responded that it did, but that its regulation was stayed under the principle of enforcement discretion.  [That FDA response unavailable; but is footnoted in newer FDA documents].  Even this early 1992 document has a context, which was a burst of FDA activity against RUO monoclonal reagents in 1991 and a proposed regulatory plan - way back then - which evolved to 1996 proposed rules for ASRs which FDA released shortly after convening an advisory panel on RUO and ASR problems (for more on 1991-1996, see here.)

2006:  More recently, after the FDA released its plans to regulate IVDMIAs in 2006 (trade press here and here and here), the Washington Legal Foundation, which has filed many FDA Citizen’s Petitions, filed a petition that the FDA couldn’t regulate LDTs.   (Here).   The FDA delayed its response to the WLF 2006 petition until July 31, 2014. [By the way, besides its own 1992 petition cited above, 14 years later Gibbs and Hyman Phelps also wrote in support of the 2006 WLF petition, here].

2008: In 2008, Genentech filed a Citizen’s Petition that the FDA can and should promptly regulate LDTs, for public safety reasons – here.

2013:  Finally, in 2013, the ACLA filed a Citizen’s Petition against FDA regulation of LDTs – here

The FDA responded to all three Citizen’s Petitions on July 31, with large parts of the three documents being the same.  

FDA Response

Not surprisingly, the FDA argued at length that it does have the authority to regulate LDTs, in response to ACLA and WLF.   In addition, the authority issue has been contested for so long that the FDA has had years to hone and polish the arguments in favor of LDT regulation authority.   It is critical that in these documents, FDA establishes both (A) its authority regulate LDTs, and (B) that its new decision to do so is rational and is not capricious: it is a reasoned, rational, and appropriately-made decision on the part of a federal agency.  That is, FDA has shown it finds itself prepared to withstand a court challenge on either (A) or (B).

The FDA argues that LDTs are devices – for example, a KRAS test based on a kit and the associated lab operations results in the same KRAS test report from an RUO- or ASR-based KRAS test brought together under LDT principles.   If the FDA can regulate the former [the KRAS kit], it must have authority to regulate the latter [an end-to-end process].   This sidesteps the point that there is a clear limitation on what FDA regulates for an IVD – which begins at the bottom of the box, and ends at the top of the box where the kit is taped shut upon manufacture.   With an IVD KRAS kit, the FDA does not regulate the practice and process of medicine after that point, just as it does not regulate off-label uses of drugs by physicians.  The FDA does not regulate or prosecute if the lab and an ordering physician uses the KRAS kit "off label" for a different cancer, or add 3 mutations to the 5 mutations packed in the IVD kit, or boil the sample for 6 rather than 5 minutes.   Nobody is fined or is sitting today in jail courtesy of the FDA for any of these lab process activities -- although people do go to jail for off label marketing of drugs and devices.   Where the FDA would begin and end its regulation of LDTs is less clear – there is no “bottom and top” of the box in the case of an LDT.  Following the FDA's analogy argument, it seems most natural that FDA regulation of LDTs would have the same beginning step and the same ending step (in the whole process) as its regulation of IVDs, if this could be defined for LDTs.

To arguments that the FDA must make changes like this by regulation, the FDA argues back with an archive of footnotes that it has always asserted its authority to regulate LDTs and the FDA has only waived the authority, year by year, so it is not asserting a new authority now.

To arguments that LDT regulation would impair access to tests and healthcare, the FDA notes that the LDT authority will be implemented gradually over many years, with checks, exceptions, and ways to speed the approval process in cases of emergencies.   

To arguments it simply lacks the statutory authority, the FDA notes that it has announced plans to regulate LDTs many times, from IVDMIA proposals to, most explicitly, its 2010 actions to convene a two day public meeting – and the Congress has never acted to amend FDA law to clarify that it couldn’t or shouldn't do so.  In fact, the FDA notes that Congress has specifically addressed the issue of the FDA regulating LDTs, by a recent statute under FDA law that doesn't restrict in any way its authority to regulate LDTs -- but only by asking the FDA to notify Congress 60 days in advance.   Such a law would make little sense if Congress already knew that FDA regulation of LDTs was impossible or illegal, and there is a legal principle against interpreting laws in ways that are null or absurd (see e.g. here). 

Interestingly, the FDA also answers Genentech that it “denies” its petition. For example, Genentech petitioned specifically for LDT regulations, but FDA responds that “regulations” are not necessary, only guidance documents.   Genentech asked that the FDA clarify a genetic test guidance document applied to IVDs and LDTs – but FDA “denies” this request, saying the guidance document contains no statements of restrictive application, so the request is unnecessary.  The FDA concludes by stating the obvious, that its current actions to regulate LDTs will largely fulfill the goals of the Genentech petition, even if the very specific list of Genentech requests are "denied."  

Note and Disclaimer (!): This webpage I’ve written is only a summary, and is not better than my understanding of some 100-plus pages of documents.  The only sure source for just what the FDA said and what it really meant is to turn to the original documentation.    


As noted by Andrew Pollack at the New York Times, the FDA was required to give Congress 8 weeks notice before releasing LDT guidance, and FDA handed the Hill this notice just as a five week recess began.

The foundational document is a “Framework” for LDT regulation (HERE).   The FDA notes that it has voluntarily allowed LDTs to be unregulated by the FDA, because they were “simple” tests primarily performed by hospitals for their own patients.   FDA notes that now LDTs may be very complicated and are often performed on a nationwide-basis by commercial firms due to the widespread access to overnight express shipping.  (I and others have made the point that just as microfluidics is one of the less-appreciated technologies for genomics, so is FedEx a technological requirement for the modern lab industry).   

Regulation will be (or, would be) implemented on a slow time table, which begins only on release of official guidance on October 1 or later, opening a public comment period.   Then, the FDA will consider public comments for up to several years before releasing final guidance (if ever).   At that point, the today’s draft document propose the following timetable:

Final Guidance (FG)

 NEW high risk LDTs must comply with premarket review.
FG + 6 months: LDT labs must comply with listing requirements.
FG + 6 months:  Compliance with adverse event reporting requirements.
FG + 12 months:  FDA begins enforcement against existing (as opposed to only new) high risk LDTs.
FG + 24 months:  Publication of a priority list for [grandfathered] high risk LDTs.
FG + 36 months:  Begin enforcement of premarket review, especially to be:  (A) Combination diagnostics (for drug prescription), (B) screening LDTs, (C) LDTs for certain infections and high risk uses including blood banking.
FG + 4 years:  Publication of a priority list for other LDTs, e.g. Class II moderate risk LDTs.
FG + 5 years:  Transition from Class III to Class II enforcement.  The FDA specifically envisions “third party review” – CAP or NY State?
FG + 9 years:  End of transition to Class II enforcement.


This section includes some exemptions which are fairly broad, and which would presumably never come into the paradigm, as it is currently laid out.  These would be subject only to “notification” so it is a safe harbor from any further regulation and review.  The FDA presents the following factors not as absolute safe harbors, but as factors considered favorably by the FDA. 

·        Safe Harbor 1:  Traditional LDTs. 
o   An LDT meets the relatively narrow definition of LDT (it doesn’t fail any of the concerns lists in “No Safe Harbor” below)
o   The LDT is used in a health care facility laboratory such as a hospital or clinic for a patient diagnosed or treated at the same facility or same healthcare system.   (See FDA regulations at 21 CFR 803.3 – the FDA already has defined “hospital system” where it defines adverse event reporting)
o   The LDT is composed ONLY of “legally marketed components” such as ASRs.   While an ASR can only be marketed as single probes, an LDT apparently may comprise more than one ASR, otherwise “component[s]” would not be plural.
o   An LDT is interpreted by a “qualified laboratory professional.”  {This is as close as the guidance gets to the concept of anatomic vs. clinical pathology.}
o   “Discretion is appropriate for LDTs available when the FDA initiated discretionary enforcement (non-enforcement) in 1976.”

Affiliation with hospitals?  Since entry into a safe harbor may be highly desired, one wonders if closer affiliation between labs and healthcare systems will occur.

Safe harbor is safer for ASRs?  The safe harbors also raise the question of whether there will be an elevated incentive to use ASR (NOT RUO) components and assemble them for your LDT.  (It appears that the lab has discretion to do what it wants with ASRs, under the ASR LDT safe harbor, but not the discretion to do what it wants in modifying an IVD).

The Hospital System safe harbor....Although certainly appealing to hospital based molecular pathologists, some may question hospital based safe harbor.  To some stakeholders, it will be unclear why a test developed by five molecular pathologists (MD) and five PhD molecular pathologists and ten M.S. support staff at a large national (and even international) reference lab would be “riskier” than the same LDT run by one MD with a thirty year old pathology credential at St. Mary’s Hospital in Podunk.  Arguably, St Mary's is exposing fewer numerical patients to risk, but the risk per patient is unchanged. Elsewhere, the FDA doesn't say devices it would claim to view as unsafe and requiring approval can be waived if used at small hospitals.

Parity of Access to Genetic Tests.   While this hospital in-network usage appeals to the FDA – it is an emblem of  FDA’s stance that formerly, tests were rarely commercial, and it reduces the need for hospitals to submit applications or FDA staff to read them -  it would create stark socioeconomic incongruities, such as rating an MGH genetic LDT safe enough for the child of a Harvard professor, but unavailable to a child with the same condition in Alabama on Medicaid (who is not being treated at MGH).     

The "Professional Interpretation" Clause.  The current text regarding “interpretation by a qualified laboratory professional” is intriguing.  Many would argue that most genetic tests require “interpretation by a qualified professional” including whole genome sequencing and broad-spectrum somatic mutation panels.  But see the original document - the interpretation exemption is for tests not using automated systems for interpretation. An automated system is used at some step in nearly every test; when does it cause the test to violate the proposed LDT safe harbors?  Is a FISH test an LDT when read with a microscope but not when read computer-assisted?

"Traditional LDTs of the 1976 Type."   The reference to pre-1976 tests is interesting, since devices introduced before and after this date is a pillar of the entire 510(k) legal system for devices and has led to a very complex and sometimes surprising body of law and regulatory practice.  If LDTs are “devices,” then this eclectic body of law and policy pivoting on the year 1976 must apply.

The following two discrete categories would also be brought under the umbrella of "notification" but not further review:

·         SAFE HARBOR 2:  LDTs for rare diseases. 
o   There are different definitions of “rare disease devices” and “orphan drugs.”  A rare disease device would be used in less than 4000 patients per year.   [If 10,000 patients would be tested to find those 4000 patients, it is not an exempt device.]

·         SAFE HARBOR 3:  LDTs for unmet needs.
o   Primarily: There is no FDA cleared or approved device
o   The device is used at a healthcare facility for its own patients, and meets the definition of LDT (manufactured and used by a single laboratory) but waiving whether it uses only "legal" components like ASRs and waiving whether it is interpreted with assistance of automation.  [Together, these laundy lists of ins-and-outs tie me in knots.]  


This section is tricky.

The FDA describes its past regulatory discretion for LDTs, and that this discretion will be unfolded only gradually as constrained by the new guidance.  However, the FDA says that the a range of tests are not (perhaps never have been) within its meaning of LDTs, even though the industry may consider the following to be LDTs.  At a minimum, the following would be subject to BOTH notification AND review by the FDA, since they are not exempt from review under a "safe harbor."

The FDA devotes considerable attention to what it has hitherto considered an LDT (I am calling this the "safe harbors" that require only notification) and what must be regulated.  The FDA now has to deal with the fact that for decades, it has said LDTs were under "enforcement discretion" but from time to time it has notified one company or another that its particular LDT was illegal.   How can this be, if LDTs were under enforcement discretion and not really defined anywhere?

Now, at this late date, it has published formal rules distinguishing between the two.  The FDA has published its rules on July 31 for what are "real LDTs" which have been under decades of enforcement discretion, discretion the FDA was apparently able to apply without defining the LDTs in the new, intricate ways.   "Real LDTs" will need only notification.   Who has been getting the enforcement letters?    Those are addressed to things that the FDA never considered LDTs (against, without published any clear list of factors).  

To me, the simplest reading is that the following are not under LDT discretion or the LDT paradigm rollout.  Under this reading, even if the FDA never finalizes the present LDT proposals, the FDA reserves the right to regulate the following at any time since they are not "real LDTs":

  • ·         An entity owns several laboratories and rolls out the test to multiple networked laboratories.
  • ·         An academic institution develops the test, and then licenses its to a private corporation with a CLIA laboratory.  “The private laboratory then manufactures the device” (aka the test, the “LDT”).
  • ·         The laboratory contracts with any third party manufacturer to manufacture a component (e.g. a gene chip plate) or “design a device” for later, final validation.
  • ·         Consider whether the LDT fails any of the criteria for a “traditional LDT” (see Safe Harbors, above).
  • ·         The test is a DTC genetic test (read the footnotes…blog dissecting this footnote, here).
There is another reading, according to Genomeweb, by which one FDA spokesperson stated by email that the what the FDA sees as out-of-bounds LDTs (those noted just above) would be regulated on just the same slow timetable as the FDA's "within bounds" LDTs.   (Here).  If so, you will be better off (at least for a few years) if the guidance is finalized than you are today, since you are not under FDA LDT discretion today, only benefiting from their lack of resources, but you would be (for a while) exempt from enforcement under the new paradigm. But 23andMe will not share in the proposed slow rollout of FDA enforcement:  while the FDA email to Genomeweb says that the various "conservative" and "non conservative" LDT tests will all be under the guidance document, a mere footnote in the guidance document says that DTC tests are under immediate enforcement.   Without this little footnote, the guidance document might have been interpreted to grant a several year reprieve to 23andMe.

Since defining these more sophisticated tests as "not LDTs" in the midst of its "LDT guidance" is confusing, this may be one area the FDA will clarify.


Under Guidance
Healthcare facility for its own patients
Longstanding and ongoing enforcement discretion

(But never before enumerated by FDA in such a clear list)
Registration only
Only “legal” components, eg ASR
Interpreted by qualified professional
LDT of pre 1976 type
Entity owns several laboratories, replicates the test across them
Subject to enforcement today, because, FDA doesn't think these are "real LDTs" at all, so, the "LDT enforcement discretion" doesn't apply to these rows today
Subject to both notification AND the 510k or PMA rules –

Unless meeting an exception such as “unmet need” or “rare” - see rows below

Potentially these test categories actually GAIN a safe harbor timetable they don’t have today (although today’s enforcement actions on these three grounds have been rare)
Academic center licenses test (as opposed to using published & unlicensed data)
Contracts with third party manufacturer (eg custom but RUO gene chip)
Rare Disease
No clear policy today
Creation of new, long term safe harbors (notification only) – thus potentially better off than today – depending on document interpretation

Expect debates over whether LDT test "X" is or is not comparable to any IVD and over the meaning of "intended use" of the IVD 
Unmet Need = No IVD
DTC Genetic Test
De facto active enforcement and warnings = cease marketing today
Same as today – no stay of enforcement – thanks to a footnote

For tests falling both in a green row and a red row - it's pretty confusing.  The green row factor will override some red row factors - but it's a mindbender.  I have an "experimental" table where I do attempt to lay out the timelines, risk factors, and exemption factors in one table - here.


The FDA sees its own “safe harbor” in arguing that CLIA does not require public reporting and filing of adverse events -- as does the FDA device paradigm.  Therefore, whether this would occur often for LDTs, FDA emphasizes this as a pillar of the LDT regulation paradigm.   Some of this material is covered within the FDA's main "Framework" document.  Readers can find more details in the full FDA guidance on this proposal in the FDA's supplementary document, here


Having just finished a biography of Lyndon Johnson - Master of the Senate - I am attuned to seeing, or imagining, stakeholder positioning.   Here are some that are near the surface.

(A) By targeting "commercial labs" with "nationally distributed tests" such as IVDMIA/MAAA tests, the FDA is exactly matching one part of the risk-based framework proposed by the College of American Pathologists, one of the largest stakeholders in this field (here).  That is, CAP has signaled several years in advance its agreement with this part of the new proposal, an apparent split with the CLIA-only position of AMP (here).

(B) By exempting hospitals and the (increasingly limitless) "hospital networks," FDA reduces the proposed workload considerably for itself while quelling the objections of hospital-based stakeholders.   This aspect of the proposal appears to favor small hospital molecular laboratories over medium or even large independent laboratories.

(C)  By focusing on LDTs more assertively as soon as an IVD is approved for an intended use, the proposal meets the longstanding and easily understood interests of AdvaMed stakeholders requesting a level playing field.  Could this backfire?  Right now, a laboriously won IVD product approval is exhausting, time consuming, and very expensive.  FDA says that many LDTs could be approved based on literature alone.  Would the IVD rather fight against an LDT copycat it can "diss" as uncertified and unreviewed - today - and will the cheaper copycat turn into a tiger if it can quickly be "FDA approved" based just on a literature file?

(D) Since the unmet need tests are defined in part by whether there is an approved IVD, versus the claim for the LDT, this could also become a battleground.  Let's say Company A has an FDA-approved test prognostic for breast cancer, and Company B has an LDT test both prognostic and predictive (of chemoresponse) in breast cancer.  Company A may point the FDA at its competitor, claiming the "unmet need" safe harbor is void.  Company B could respond that since its' test is predictive, it fulfills an unmet need no IVD can match.   FDA would have the dilemma of determining whether Company B actually is predictive - something that normally takes extensive and elaborate technical review; or find another way to resolve the tempest in the safe harbor.  Similarly, does a 5-mutation LDT KRAS test fulfill the same purpose as a 3-mutation IVD KRAS test: is the 5 mutation LDT in a safe harbor under "unmet need?"   What if the FDA isn't sure mutations 4 and 5 are useful?  What if a lab adds mutations just to create a safe harbor?  Who decides what's legit?  How long does it take?  Who pays?

(E) The proposals do not say very much about two aspects that have attracted concerns, the use of quality systems reporting (QSR, here and here) and the use and enforcement of "labeling."   Regarding labeling, if a lab sets up a Her2Neu test as a "prognostic" test (lower risk category) but a clinician uses it for drug selection (higher risk category), what is the role of FDA?   FDA product labels are often the result of months of logrolling and parsing alternatives (see the complex ProSigna Nanostring label, here - including statements such as "not intended therapy for patients").  How would this lengthy and complex labeling negotiation be rolled out for LDTs?  To ensure only on-label claims are made by the "manufacturer," how would this be enforced for hospital systems where laboratory staff and clinicians are all under the same employer?   Could the same pathologist make only "on-label claims" for their LDT KRAS test in his lab office, but make general subject matter expert statements just down the hall while he is at tumor board?

(F) The "Where" and the "Parts" - Tests like next generation sequencing are uniquely suited to operations in different places - the DNA extraction could be performed in one place, the wet chemistry in another, the assembly in a third, and the bioinformatics and interpretation in a fourth. How this would be accommodated under the FDA regulation - or for that matter under Medicare rules, such as "only the performing lab may bill" and "date of service" - is unknown.  If an industry emerges to produce modular parts of tests like this, it could promote expertise, efficiency, and innovation.  How does it fit with a 1990s version of LDT regulation?  This would be a category of innovation - modularity - that might be stifled under the FDA regulations and no one would really know, because it just wouldn't exist.

(G) Finally, one can take a purely administrative view of the FDA - even if this proposal is never implemented, and something someday goes far wrong with an LDT, no one can say the FDA didn't try and do so with a full-court press (see, e.g. Carpenter's FDA book on internal FDA politics, Reputation and Power).

Show Me The Money

I am not an expert on FDA fees, and their are many ins-and-outs such as small business discounts. But pages on 510(k) fees are here, and PMA fees here.  Some screen shots are:

And for manufacturer [factory] registration and listing fees (here):

Some Additional Weblinks:

FDA Press release:

CDRH Chief - Blog on the Announcements
Genetic Engineering News

MedPage Today

AMP Voices Concern - Fierce Diagnostics
Congressman Burgess Expresses Concern over Guidance
And finally.... I was discussing this with a colleague, normally the FDA regulates drugs and devices.  So, there's a "box."   The FDA regulates "what's in the box."   If you regulate an LDT, where's the box?  Where are the borders of what the lab does that the FDA (by the FDA's analogy to an IVD) could regulate, versus what is clearly only the provenance of medical (healthcare) practice?   I was reminded of Clara Peller, the 1984 commercials, 'Where's the Beef?"  (here).   So the modern day update would be: