Thursday, October 2, 2014

FDA and LDT’s: Should it be "Guidance" or "Regulation"?

Since a public meeting in 2010, the FDA has been discussing its plans to bring laboratory tests at local hospitals and laboratories under its scope of regulation.   This activity escalated quickly with the release of a planned regulatory framework to Congress on July 31, 2014, and the official release of the same documents on October 3, 2014, kicking off a 120 day public comment period (here).

On September 9 (here), the House Ways & Means committee held a hearing on the proposed new scope of FDA activities.  In that hearing, some lawmakers and public stakeholders repeatedly suggested that the FDA should undertake these activities only after formal federal notice & comment rulemaking.   But Dr. Shuren of the FDA repeatedly emphasized that the FDA had solid legal grounds allowing it to regulate laboratory developed tests by issuing and finalizing a “guidance document.”  

What would really be gained by the rulemaking process?   
  1. For one thing, for regulations proposed that would have a greater than $100M economic impact, a formal budgetary analysis must be made by the agency and reviewed by the Office of Management and Budget.   
  2. For another, new regulations (law made by an agency and codified in the Code of Federal Regulations) would be proposed, open to comment, and then finalized verbatim and rarely changed.  
  3. Finally, policies that are related to, but not literally laid down in the C.F.R. regulation would also be presented  and generally followed by the agency until changed through another round of agency proposal and public comment.  

We discuss the FDA LDT proposals from this perspective after the break, borrowing some analogies from CMS regulations and policy and from the FDA formal rulemaking for Analyte Specific Reagents (ASRs) in 1996.

The Administrative Procedures Act of 1946 requires federal agencies to act in a rationale and publicly open way in setting rules and policies that are within the scope of their enabling statute.   Typically, this involves proposing regulations (agency-written law) for public review, with an extensive discussion of the rationale behind the regulations.  A few months or up to two years after the comment period, the regulations are finalized.   The final regulations are put in a codebook called the Code of Federal Regulations (CFR) – for example, FDA regulations are in Section 21 and CMS regulations are in Section 42.  For the CFR, see online here; Wikipedia here.

In the case of laboratory tests, the FDA argues that it has authority to regulate medical devices dating back to the 1976 medical device act.   It argues it has only voluntarily held itself back from acting on its authority, and details at great length in its new proposed guidance documents why it now can and should regulate LDTs, and provides a general framework for how it will do so.  The FDA’s attorneys have studied the matter and are certain that formal “regulation” is not necessary.

The Wide Gap between Regulation and Practice:  
    The example of the Physician Fee Schedule

Physicians are paid under Medicare Part B under a fee schedule of daunting scale, detail, and policy complexity.   There are likely only a few people in the United States that fully understand how surveys, prices, inputs, rules, calculations, and procedures inside CMS computers product the exact prices of physician services.   The Physician Fee Schedule concept comes from the Statute – the Social Security Act, Section 1848.   In key and concise parts, the statute states:

(1) DIVISION OF PHYSICIANS’ SERVICES INTO COMPONENTS.—In this section, with respect to a physicians’ service:
(A) WORK COMPONENT DEFINED.—The term “work component” means the portion of the resources used in furnishing the service that reflects physician time and intensity in furnishing the service. Such portion shall—(i) include activities before and after direct patient contact, and (ii) be defined, with respect to surgical procedures, to reflect a global definition including pre-operative and post-operative physicians’ services.
(B) PRACTICE EXPENSE COMPONENT DEFINED.—The term “practice expense component” means the portion of the resources used in furnishing the service that reflects the general categories of expenses (such as office rent and wages of personnel, but excluding malpractice expenses) comprising practice expenses.
COMBINATION OF UNITS FOR COMPONENTS.—The Secretary shall develop a methodology for combining the work, practice expense, and malpractice relative value units, determined under subparagraph (C), for each service in a manner to produce a single relative value for that service.

These key sections of statute are only 169 words long.  They are repeated without much more detail in CMS regulations at 42 CFR 414.22

CMS establishes RVUs for physicians' work, practice expense, and malpractice insurance.  Key passages are about 260 words long:

(a) Physician work RVUs—
(1) General rule. Physician work RVUs are established using a relative value scale in which the value of physician work for a particular service is rated relative to the value of work for other physician services.
(2) Special RVUs for anesthesia and radiology services)—
(i) Anesthesia services. The rules for determining RVUs for anesthesia services are set forth in §414.46.
(ii) Radiology services. CMS bases the RVUs for all radiology services on the relative value scale developed under section 1834(b)(1)(A) of the Act, with appropriate modifications to ensure that the RVUs established for radiology services that are similar or related to other physician services are consistent with the RVUs established for those similar or related services.
(b) Practice expense RVUs.
(1) Practice expense RVUs are computed for each service or class of service by applying average historical practice cost percentages to the estimated average allowed charge during the 1991 base period.
(2) The average practice expense percentage for a service or class of services is computed as follows:
(i) Multiply the average practice expense percentage for each specialty by the proportion of a particular service or class of service performed by that specialty.
(ii) Add the products for all specialties.
….For services furnished in 2002 and subsequent years, the practice expense RVUs are based entirely on relative practice expense resources…..For services furnished in 2002 and subsequent years, the practice expense RVUs are based entirely on relative practice expense resources….CMS establishes criteria for supplemental surveys regarding specialty practice expenses submitted to CMS that may be used in determining practice expense RVUs..

The point is that the description of the RVU process in statute and regulation is only a few sentences and leaves a great deal to agency discretion.  

Most of the mountain of rules used to create RVUs are not found in the statute or the regulation – they are just “agency policy.”      We look in vain in the terse statute or regulation, trying to capture what physicians, policy experts, and journalist find important about RVUs (see a September 2014 report, here.)  The statute and regulations tell us nothing about whether the values are "right" or not - what policymakers and the public worry about most.  The dry words of the short regulations quoted above don't capture what observers find either inspiring, horrifying, or mind-blowing about the RVU system.

Could FDA “regulations” for LDTs similarly be written, be a few hundred words long, and not capture at all  what people find most crucial about the FDA LDT process?   

Let’s look at a parallel that is within the FDA – the production of Analyte Specific Reagent regulations in 1996.

The History Before Analyte Specific Reagent Regulations

This history is so old I’ve only been able to piece parts of it together by internet research.   For example, a 1997 article by Malinowski (here) has footnote 113, which states that a number of diagnostic blood tests available in Europe are not available in the US, “Part of this is the result of Centocor’s experience with their test for CA 15.3, a marker for breast cancer.  In 1991, the FDA forced Centocor to stop selling this test as a research product in the United States.”   I tried pulling this thread, and came up with a 1992 article in Medical Laboratory Observer (here), “Threats to Home Brew Products Anger Labs,” and oddly enough, here, in a 1993 Journal of Insurance Medicine meeting summary.  

According to Journal of Insurance Medicine (1993),

The FDA in 1991 “circulated a draft compliance policy to several manufacturers for their informal comments.  The FDA stated…numerous laboratories have been manufacturing “home brew products…and utilizing these for unapproved diagnostic purposes.”  FDA added, “home brew products are subject to the same regulatory requirements as any unapproved medical device.”  The Journal says, “The [draft] created a reaction of confusion due to the profound adverse effects of the proposed policy… the FDA draft guide would interfere with CLIA 1998 authority…”  

Medical Laboratory Observer  wrote (1992):

 A THREAT to clinical laboratories' use of so-called "home brew" testing products may never be carried out. The ruckus raised by a draft policy guide, however, is an indication of just how much some Federal regulators still have to learn about common industry practices.

The saga began last year, when officials at the Food and Drug Administration (FDA) said investigations found that a manufacturer was commercially marketing devices approved for research use only. [Presumably Centocor per other references]

The agency warned the firm that continued distribution would lead to regulatory action, and labs were notified that some of the company's diagnostic products would no longer be available.

That incident set off an alarm at the College of American Pathologists (CAP), where officials worried that a more widespread policy was in the making. True to those concerns, the FDA sent an October 1991 warning to six other manufacturers of in vitro diagnostic products on the implications of promoting tests for unapproved uses.

At the FDA's request, in February CAP submitted a list of 43 monoclonal antibodies that CAP said should be allowed to stay on the market since they had become the standard of care for diagnosing certain diseases. According to CAP, many of these products are currently approved as serum markers in flow cytometry or immunocytochemistry….

The 1997 final rule makes a brief reference to a previously released 1992 document titled, "Commercialization of Unapproved IVDs Labeled for Research and Investigation." (62 Fed Reg 62251, also mentioned in FN 7 of the 2013 ACLA Citizen's Petition against LDT regulation).

The history just related about the unfolding of 1991 and 1992 events also provides a backdrop for a Citizens Petition 22 years ago in 1992, that FDA should not regulate LDTs (see the "1992" section, here.) [*]

The History of Producing Analyte Specific Reagent Regulations

                On January 2, 1996, the FDA announced a public meeting on the topic of “analyte specific reagents,” to be held on January 22, 1996, and requesting public comments prior to January 8, 1996 (61 Fed Reg 74-75).    Although a transcript of the committee meeting is not available online, results of the meeting were discussed in proposed regulations for ASRs which appeared on March 14, 1996 (61 Fed Reg 10484).  The full eight page proposal and regulation, about 6000 words long, of which are 1200 are the actual ASR regulations.  The whole proposal from March 1996 is available online here; I quote an introductory section:

FDA currently regulates the safety and effectiveness of diagnostic tests
that are traditionally manufactured and commercially marketed as
finished products. However, in-house developed tests have not been
actively regulated by the Agency and the ingredients used in them
generally are not produced under FDA assured manufacturing quality
control. Other general controls also have not been applied routinely to
these products.
 FDA is not proposing a comprehensive regulatory scheme
over the final tests produced by these laboratories and is focusing
instead on the ``active ingredients'' (ASR's) provided to the
laboratories. However, at a future date, the agency may reevaluate
whether additional controls over the in-house tests developed by such
laboratories may be needed to provide an appropriate level of consumer
 Such controls may be especially relevant as testing for the
presence of genes associated with cancer or dementing diseases becomes
more widely available. Additional controls might include a broad array
of approaches, ranging from full premarket review by FDA to use of
third parties to evaluate analytical or clinical performance of the
 The laboratories producing tests from ASR's and offering the
tests as laboratory services are currently regulated by the Health Care
Financing Administration (HCFA) under the Clinical Laboratory
Improvement Amendments of 1988 (CLIA-88) for compliance with general
laboratory standards regarding personnel, proficiency testing, quality
control, and quality assurance.
 However, these HCFA regulations do not
include the same product controls provided by FDA. As a result, neither
patients nor practitioners have assurance that all ingredients in the
laboratory developed tests are of high quality and capable of producing
consistent results.
  FDA is concerned that the present situation with respect to in-
house developed tests, in which these ingredients are essentially
unregulated and therefore of unpredictable quality, may create a risk
to the public health.

The regulations are in three parts, as proposed, 21 CFR 809.10 has to do with labeling of ASRs (new section [e]).   Section 809.30 restricts the sale of ASRs, and section 864.4020 defines ASRs generally as Class I but elevates blood supply ASRs to Class III.   (It’s difficult to imagine how you would produce a PMA on an ASR for blood reagents, since it is not a complete test.  The FDA says on page 62257 that Class III ASRs would only be reviewed “with the test that uses it,” i.e. the downstream finished LDT).

The final regulations on ASRs were released on November 21, 1997 (62 Fed Reg 62243, here.  One example of public comment urging tougher ASR regs, here.)  The actual regulations had minimal change; with public comment the document was now 19,000 words instead of 6000.   The point is - 

As many readers may remember, about 7-10 years ago, the FDA became very concerned in the decade following  the 1997, manufacturers were skirting the FDA’s vision of how limited an ASR is (it could not be 3 ASRs in one box, for example), and the FDA released additional guidance on ASRs in a Q&A format in 2007 (here).      

The ASR Regulation of 1997:  
How It Described Its Own "Economic Impact"

The economic impact of the rule is administrative only.  For example, if you spend 20 hours complying with the rule, stop sales, lose $50M in revenue, and go bankrupt, the economic impact of the rules is 20 hours times, say $30 per hour or $600.  

In the ASR rule, the economic description begins on page 62256 (62 Fed Reg 62256).   The FDA framed its economic analysis in two ways:

Manufacturers are required to:
Factors that Scale the Work
1) Register and list ASRs with the agency;
2) Conform to medical device good manufacturing;
3) Comply with medical device reporting;
4) Relabel products;
5) Restrict the sale of ASRs to clinical laboratories performing high complexity testing.   
1) Whether the firm makes FDA regulated products already;
2) The nature and number of its ASRs;
3) The size of the firm;
4) The adequacy of in-place quality procedures.

The then walked through these factors one at a time.  The FDA estimated that registration and listing would involve 1-2 hours of time for up to 300 manufacturers, costing $5000-10,000 in total per year.    GMP compliance would be minimal for firms making GMP products already, but could cost $50,000-$200,000 for companies that have never implemented GMP processes.   “FDA cannot estimate the economic impact because the agency does not know how many [ASR firms] also make regulated medical devices.”   The agency estimated the cost of each label redesign at $89 or about $1M for 10,000 ASRs.   The restriction on sales of ASRs, “The agencies believes this restriction will have no economic impact…received no comments that suggested otherwise.”   

The FDA also estimated the impact on clinical laboratories would be minimal, except for a one time $80 cost to add software to add the disclosure statement about ASR usage to its laboratory final reports. 

While acknowledging there were many unknowns in its analysis, FDA concluded the implementation of the ASR rule would be “far below the $100M threshold for an economically significant regulation.”  For example, even with 15,000 hours of effort (table) at $50/hr, it would be $750,000.

What Changed As A Result of Public Comment?

The simplest answer is that the 18 months review and many pages of public comment resulted in few changes in the final regulation.   There are wordsmithing and renumbering changes, for example, the word "ligand" is added to the definition of an ASR.  The final regulation has a carve out for VA labs that were not under CLIA.   The most substantive changes are two:

  1. Modest revision of Class I, II, III framework.  FDA originally proposed that simple ASRs would be Class I, but ASRs used in potentially fatal infectious disease and blood banking would be Class III.   However, ASRs would be "saved" from Class III when intended for a use that was already approved for a Class II IVD.   The final regulation explicitly carves out categories for Class II and Class III ASRs.  These are oddball ideas, however, since the "ASR" has to be approved as "Class II or Class III" only when used in a named approved test, which would be an LDT.  If it were used by inclusion in a completed IVD, it wouldn't be an ASR anymore, just a manufacturing supply.  
  2. Labs must add disclaimers to all lab reports.  The proposed regulations did not include a section requiring labs to place a disclaimer on every ASR based lab report that the test used was not FDA approved.   However, FDA did solicit comments on whether this was a good idea in the proposed rulemaking, relying heavily on the fact this was an idea voiced by its external advisory committee.   Thus, the final regulation and only the final version of the regulation displays the remarkable section 809.30(e), as part of the "restrictions on sales," that only a licensed physician can order an ASR based test and furthermore, the lab report must in every case include a disclaimer the test result is ASR based and not FDA approved.   This is a remarkable reach-through between FDA law and LDT and CLIA law.  

What Was Accomplished by Rulemaking?

It was about 18 months between the proposed and final rule, and the FDA relied heavily in its discussions on a public advisory committee held two months before the initial rulemaking.  Arguably, the ASR regulation was slowed down for 18 months.  However, it's not clear it "had to be" that long: CMS produces rule changes of similar complexity every year for Part A and Part B in a five month comment period (e.g. July 1 proposed, November 1 final.)  FDA routinely takes 12-24 months to finalize guidance, just as it took 18 months to finalize this rule.

Rulemaking required FDA to publicize its budget estimations of purely administrative (not sales or economic) impact.  But in some cases, the FDA could stay, "it is impossible to know [X] or [Y]," so the economics could not be estimated.  

FDA efforts on ASRs stretched from 1990/1991 to proposed regulations in 1996 and final regulations in 1997.   FDA current efforts on LDTs range from IVDMIA guidance in 2006 to public meetings in 2010 to draft guidance in 2014 (albeit probably written by 2012).  The timelines are surprisingly similar.

ASR vs IVD issues remain alive in 2014.  For example, the FDA held a 2013 conference on flow cytometry issues, which revolve in part around ASR and non ASR reagents.   In October 2014, the FDA released a new draft guidance on this topic for flow cytometry (here).


[*] For historical context to these 1991 and 1996 ASR events, PSA was approved for cancer monitoring in 1986 and for screening in 1994.  FDA was likely concerned that some immunoassays were passing through FDA approval and other contemporary RUO ones were being used as LDTs.