Wednesday, January 2, 2019

Has Medicare Canceled Much of Germline BRCA Testing, Including for Ashkenazi Populations? New 2019 MolDx LCD

Today I was scanning 2019 LCDs for updates and found an unusual one in the MolDx program.

Search the Medicare Coverage Database for the BRCA panel code 81432.  You'll get quite a set of LCDs:

Click on the L36082 LCD for BRCA/BRCA2 genetic testing, which are frequently performed by NGS sequencing or panel testing.   Find the Palmetto LCD L36082 at this link.  The new text is clipped for you here:

Multigene Panels***

^ The indications and limitations of coverage listed in National Coverage Determination (NCD) 90.2 (Next Generation Sequencing- NGS) apply to genetic testing for susceptibility to breast or ovarian cancer. While the NGS NCD Section 90.2 B describes specific coverage criteria for nationally covered tests, Section 90.2 D permits coverage of other NGS as a diagnostic laboratory test for patients with cancer when performed and ordered according to the requirements described by the NCD. According to Section D of the NGS NCD AB Medicare Administrative Contractors (AB MACs) may cover next generation sequencing tests in patients with cancer. As such, genetic testing for susceptibility to breast or ovarian cancer with multi­-gene NGS panels (not otherwise covered under NCD 90.2 Section B) may be covered by this AB MAC as reasonable and necessary when ALL of the NCD criteria are met in addition to the following... (next section unchanged)

This referral to "coverage requirements described in the NCD...ALL of the NCD criteria are met..." acts to allow NGS based testing only in women with Stage III/IV cancer, excluding women with Stage I/II cancer.   See the CMS NGS NCD here.  A November 30 transmittal includes the text, "A diagnostic laboratory test using NGS is non-covered when cancer patients do not have the above-noted indications for cancer."  This makes it sound like even, for example, NGS based infectious disease testing is blocked - "a test using NGS."

According to, 60% of breast cancer cases are localized, e.g. stage 1/2, suggesting the rule if applied this way will impact a lot of testing.

The NCD, somewhat bizarrely, would act to block NGS method-based testing but allow sequencing and reporting of the same genes, in the same Stage 1/2 women, under the LCD, with Sanger or PCR testing. 

BRCA and Ashkenazi Populations

BRCA mutations are far more common in Ashkenazi-heritage populations.  Brandt-Rauf and colleagues write, "Lack of awareness about the possibility of hereditary breast cancer in the Jewish population which . . . has enormous risk compared with other populations—maybe ten-fold higher of having a mutation—makes it rather disadvantageous to not talk about it...The truth demographically is that over 90% of Jews in North America are of Ashkenazi origin."

Similarly, in 2017 CDC reports that "Women who had Ashkenazi panel testing accounted for approximately 5% of women who had any BRCA testing during the study period and ≤0.1% of all women aged 18–64 years."  This was age 18-64 related to employer preventive benefits under ACA; rates >65 are similar.

The USPSTF also reported about a 10X increase in prevalence of BRCA carriers in the Ashekenazi population, 2013, here


As of January 2, 2019, the same CMS database did not list this change in the MolDx LCD for Noridian Jurisdiction E or F.   (Those LCDs haven't been updated for routine new 2019 BRCA codes either, though).  See here.

There are several reasons why it hasn't made sense, to me, to apply the March 2018 NCD to germline testing like this.   (And no MAC has come up with this explicit application to germline testing until this week, showing it is not an obvious reading of the full NCD.)

  1. The NCD was based on an application by Foundation Medicine for FDA-approved CDx tumor testing.  All the literature and discussion is on tumor testing.  Targeted drugs are nearly always used only in metastatic disease, where the Stage III/IV rule makes sense.  That rule makes no sense applied to germline testing, and germline testing wasn't discussed in the NCD.  Germline testing couldn't even get into the main NCD coverage rule, since it isn't a CDx.   It seems more like a semantic unintended effect, given the whole body of evidence.
  2. In response to concerns from several stakeholders that the NCD (in draft form) might be misapplied to germline testing, CMS intentionally inserted explicit new text directly into the body of the NCD that it can be applied only to a certain type of NGS testing, for targeted therapies.  This would seem to exclude application of the rules to germline testing and narrow them only to targeted therapy testing.  (I wrote about this on April 18, 2018).  
    1. CMS also states in the NCD, "BRCA is discussed in familial risk assessment (Daly et al. 2017) which is outside the scope of this decision."
    2. Daly 2017 means NCCN guidelines for hereditary testing, which have a special section for patients with personal history of cancer, the section LCDs provide coverage for in all 50 states.   CMS proactively negotiated this position for MACs around 2005 - that BRCA testing was covered (with conditions) for Medicare patients with a personal history of cancer.  It makes no sense to revoke this selectively and solely for NGS method testing.
    3. CMS also left all germline guidelines (e.g. Daly NCCN) out of its Guideline section, but included somatic testing guidelines.
  3. NCDs are based on the Medicare statute for reasonable & necessary medical care.  It makes no sense the exact same mutation would be legally medically necessary (under the same CPT code and cost) if found by one method rather than another.  Here, CMS would be saying that it is legal and medically necessary to pay $2000 for 81162 BRCA testing by Sanger method, but illegal and medically unnecessary to pay exactly the same $2000 for same code 81162 BRCA testing by NGS in the same patient.   There can't possibly be any rationale there, given the identical results and identical cost to CMS.


In addition to the links above, in the cloud I've put a PDF copy of the Palmetto January 2019 LCD, and a copy of a Word-based redline of recent changes.  Here and here, respectively.

Additional Considerations

Shift to Sanger Testing for Ridiculous Reasons?

Nothing in the new LCD prevents gene testing by non-NGS methods in Stage 1-2 patients.  To the extent NGS testing for germline risk mutations is allowed only in Stage 3-4 patients, this is probably counterproductive since most benefit for BRCA-adherent longitudinal management is going to be in the longest-surviving patients, not those terminally ill or entering hospice care with Stage 4 cancer.

In addition, while most BRCA testing would be equal dollars under the NCD (e.g. 81162 for BRCA sequencing and dup deletion costs no more or less to CMS with either method) - some costly loopholes are opened up.  For example, every MAC except the NGS MAC (ironic name) covers 81432, a ten-gene panel for breast cancer.  MACs aren't covering "the code," they are signally they are covering all ten genes, including BRCA1/2.  This means if a lab performs 9 genes, they are separately billable.  Here, you hit an NCCI new rule that you aren't supposed to stack Category I genes with NGS methods, but use 81479 instead.   However, that NCCI rule doesn't apply to Sanger or PCR, so it seems you CAN stack the nine HBOC genes, and starting with $2000-plus for BRCA.

MolDx LCD Instruction Is Ambiguous

The new LCD instruction tells people doing HBOC testing they must refer to the NGS NCD if they are using NGS methods.

But they don't tell you how to interpret the NCD when you get there.  Do you follow the sentence that says only recurrent/metastatic cancers can be covered, or, do you follow the sentence in the NCD that says the NCD only applies to one subset of NGS testing, the kind for targeted chemotherapy? 

What If Your Stage I Breast Cancer Is Gone?

The NCD says that for patients with cancer, NGS testing can be covered by LCDs only if the patient has advanced cancer.  However, a patient who has stage 1 breast cancer and a lumpectomy no longer has any cancer - no signs and symptoms of cancer.  She had cancer before the lumpectomy that removed the stage I cancer.  So does the NGS NCD apply to her - no longer an obvious cancer patient - or does the LCD concept apply (a person with "a personal history of cancer," which she does have?)

Are Medicare Advantage Plans at Compliance Risk?

Medicare Advantage plans can cover services non-covered by CMS, like vision, dental, or in this case, stage 1/2 HBOC testing with NGS.  However, Medicare Advantage plans must meet compliance rules and regulations when they cover non-covered services, like separate notification to the patient in advertising materials and separate budgeting reporting to CMS.  Have Medicare Advantage plans been legally out of compliance since March 18, 2018, for reporting Stage 1/2 HBOC testing within their covered costs under Medicare?

Interpreting the NCD Ambiguity to Avoid, Not Create, Absurdity

There is a tsrong concept in law that regulations should be interpreted, if possible, in a way they make sense.  One sentence of the NCD says that NGS testing can only be covered in cancer patients if they have advanced cancer.  Another sentence says that the NCD as a whole applies only to a subclass of NGS tests for targeted therapies.   The only way to coherently interpret these is to assume the second sentence modifies the meaning of the first.   That's logical.  The other way around doesn't work.
  • See Justice Scalia's book on canons of interpretation, here.
  • See an 1892 Supreme Court case with paragraphs of discussion of the doctrine that interpretations of a law must avoid absurdity here.  
  • See a 2002 article by Cass Sunstein here.  
  • The second SCOTUS ACA case, King v Burwell 2015, also hinged in part on avoiding an absurd interpretation of the meaning of "state"; here.
  • On interpreting laws into practice in ways that are not invalid or nonsensical, see also Ginsburg 1979 here and Levy v Louisiana here.
Which Is Controlling?  The Header Coverage Section of the NCD or the Sentence Restricting Its Scope to Targeted Therapy Testing?

I've been told that CMS argues that the "header section" of the NCD is the operative section, and nothing in the body can modify the header section.  (The header section says only advanced cancer patients can be tested by NGS, but the body says the scope of the NCD is only targeted therapy testing.) 

Let's assume that only the header section is the "operative" section.  Where is that written down?   Is that in the statute?  Is that in any regulation?  Is that even in the CMS manuals - anywhere?   Maybe it's just a subregulatory concept at CMS.  But if a subregulatory concept can modify the interpretation of an NCD, then why can't a written statement in the body of the NCD modify the NCD?  Even the Supreme Court frequently uses the framing of a law as a whole in order to interpret it.

CMS Subregulatory Language Often Modifies Regulatory Language

Even if you argue the header of the NCD has the status of "regulatory language" and the body of the NCD the status of "subregulatory language," the latter frequently modifies for the former. 

For one solid example, in the 14-day-rule, the regulatory language (42 CFR 414.500ff) states parts of the rule apply to molecular pathology tests.  You can't tell from the term "molecular pathology tests" if this applies to molecular microbiology or not.  You have to look to the context and explanations in subregulatory rulemaking to find it only applies to human molecular tests - you can't tell from the regulation, and in fact, the regulation might more easily imply the opposite.  (Wikipedia: Molecular pathology is commonly used in diagnosis of cancer and infectious diseases."

Similarly, the regulation about a recent required exception to the 14 day rule (for genomic tests) took effect January 1, 2018, and that regulatory text hasn't changed.  However, subregulatory (PDF) announcements from CMS have extended implementation, most recently updated on December 26, 2018, until July 1, 2019.   CMS regulations would fall apart and be unusable, resulting in chaos, if subregulatory definitions and clarifications did not exist and couldn't modify regulatory language.

Press Release Makes No Sense

In its press release on the NCD, CMS said that "advanced diagnostic laboratory tests now have expanded Medicare coverage."  This would not be true; FMI test already had coverage as did the Oncomine FDA test.   There was no expansion.  Under the currently floated interpretation, there would be only a retraction, of coverage under LCDs in 50 states for certain specific germline testing.  In 2017, CMS paid only $7M in 81445/81450/81455 testing, hardly meriting an NCD in a $300B program, but CMS paid $75M for germline testing (various 2017 BRCA codes), of which about 2/3 would be in stage 1/2 patients, and almost all by NGS, so retracting $50M of coverage.

CMS Position Violates Statute in a Catch-22

Here's what statute says about NCCN process (at 1862(a))

In making a national coverage determination (as defined in paragraph (1)(B) of section 1869(f)) the Secretary shall ensure consistent with subsection (l) that the public is afforded notice and opportunity to comment prior to implementation by the Secretary of the determination; meetings of advisory committees with respect to the determination are made on the record; in making the determination, the Secretary has considered applicable information (including clinical experience and medical, technical, and scientific evidence) with respect to the subject matter of the determination; and in the determination, provide a clear statement of the basis for the determination (including responses to comments received from the public), the assumptions underlying that basis, and make available to the public the data (other than proprietary data) considered in making the determination.

1) If CMS intended to exclude germline and infectious disease testing, and carefully reviewed all the appropriate  guidelines and data, it violates statute here, because it didn't list for the public the materials reviewed on those topics or its reasoning on those topics.  In fact it did the opposite - saying things like NCCN guidelines Daly 2017 were not reviewed and out of scope.

2) If CMS intended to exclude germline and infectious disease testing, but didn't carefully review any materials or guidelines, then it violates statute by issuing a decision with no justification.   Bang.

A FOIA request or an NCD Appeal (SSA 1869(f)(1), with discovery of documents) could distinguish between #1 and #2, but neither is allowed by the statute quoted.

3) If CMS didn't really think ahead to implications of some of its language choices, and other sections of NCD indicate other intentions, then CMS doesn't have to block germline testing, under the avoidance of absurdity rules cited earlier in this blog.

Another option related to these ideas is that CMS intended to block infectious disease and germline testing, but viewed any and all prior textbook knowledge (such as the (i) discovery of clinically useful bacterial resistance genes or (ii) the discovery of BRCA founder mutations like 185delAG) by PCR or Sanger methods as irrelevant to assessing any value of clinical reporting of that clinical mutation by NGS, and thus, such PCR or Sanger literature is out of scope, accounting for the absent bibliography.  (But this, too, is a trip to Alice in Wonderland).

CMS Could Not Possibly Have Been Thinking About Non-CDx Tests In Writing Proposed NCD in November 2017

Internal evidence (IN THE HEADER SECTION) makes clear that CMS could not possibly have been thinking of all types of NGS testing when writing the proposed NCD, the NCD that the public had an opportunity to comment on.

Recall that the original NCD proposal covered FDA CDx tests on label, and covered NGS testing off label only in two conditions.  One, was NGS testing in advanced cancer with an FDA approved CDx but used off label.  This required an elaborate registry.  Second, was NGS testing not FDA approved (e.g. LDT) which was covered only in an NCI trial.

Come forward to the NCD Transmittal, which includes a statement that all NGS testing not in advanced cancer is not covered.   If the original NCD really was meant to apply to all uses of NGS testing methods, why did it require RECIST testing, instructions on comparing results to "initial clinical validation of the companion diagnostic"?   Obviously, these requirements make sense ONLY if the author was only thinking of CDx uses of NGS in cancer, and not other uses like germline or infectious disease testing, in which case, these rules are just nonsense and cannot be applied.  You cannot compare the results of NGS infectious testing "to the original companion diagnostic," yet this would have been required.  On the other hand, if CMS came up with the idea to apply the NCD to all types of testing for whatever purpose if NGS is involved, only in the final draft, there was no public opportunity to comment on that idea.

Oliver Elemento's Op Ed About NCD in WSJ (March 2018)

Dr. Elemento raised numerous concerns about unintended effects of the NCD in WSJ, March 2018, here.

Typo in the NCD.  

There is an apparent typo in the Q&A section of the NCD.

Comment: A few commenters suggested that a clarification or discussion of testing and reporting of germline and somatic changes be explicit as a requirement for the report of test results in the final NCD.

Response: The final decision allows laboratories and diagnostic laboratory test developers to determine what is included in the test report. Further, the final decision is unchanged from the proposed decision in that it does not exclude coverage for testing of somatic mutations, provided the other coverage criteria are met. We note germline testing in the absence of signs or symptoms is designated a screening test and not a diagnostic laboratory test, which remains outside the scope of this NCD.

The sentence that "germline testing in the absence of signs and symptoms" is a screening test; that is nothing new and not relevant to the germline NCDs.  Above that, CMS answers that the NCD "does not exclude coveage for testing of somatic mutations."  Of course it does not; it never did, it couldn't, and no one ever asked if the NCD on the FMI test excluded reporting of any somatic mutations.   That phrase only makes any sense if CMS meant "does not exclude coverage for testing of ^germline mutations" meaning here in addition to somatic mutations.