Tuesday, January 22, 2019

Genomeweb Publishes Article on CMS's New NGS NCD Interpretation

At Genomeweb, journalist Turna Ray publishes a long-form article on the CMS NCD for NGS testing in advanced cancer, January 22, 2019 - subscription, here

As revealed in January 1 edits to a MolDx LCD for hereditary breast cancer testing in patients with a personal history of cancer, the NCD is being applied to non-tumor testing based on a sentence in a November 2018 CMS transmittal to its contractors.  That transmittal remarks, "A diagnostic laboratory test using NGS is non-covered when cancer patients do not have the above-noted indications for cancer under either national or local coverage criteria."

Read literally, this ropes in NGS tests that are otherwise entirely outside the scope of the NCD, such as infectious disease or hereditary testing - any medical test utilizing a step that involves NGS.   


What Happened?

In November 2017, CMS proposed coverage of the Foundation Medicine F1 CDx test on-label, with off-label indications covered under CED that compared outcomes to on-label combination diagnostics.  The NCD proposal was clearly about CDx testing; the only way you could get coverage, even under CED, was as a CDx tumor test.  In CED, tests like RECIST testing for tumor size response to a targeted drug were required.

In March 2018, CMS provided final coverage to the F1 CDx test, and allowed LDT NGS tumor tests to be covered in advanced (stage 3-4) cancers.   The NCD's didn't discuss infectious disease testing with NGS or hereditary testing with NGS, except for a remark that hereditary testing (e.g. NCCN guidelines Daly et al., on which all LCDs are based) were out of scope of the discussion.

Procedural History

The procedural history actually goes back to 2004-2006.  For a short period the CMS contractor for Utah covered BRCA testing in high-risk women with a heavily loaded family history as well as women with a personal history of breast cancer.  In June 2006, working with the CMS coverage group, a new LCD was designed that covered BRCA testing only in women with  personal history of cancer - e.g. management of their ongoing disease.   (I was working for a carrier at the time, so I knew the story at the time.  Dr. Alton Wagnon was the Utah medical director.  A public article at FORCE from the period is here.  The resulting LCD is here in its 2008 version.)

So CMS Coverage Group specifically endorsed criteria (quite similar to today's NCCN criteria Daly et al.) for BRCA testing in women with a personal history of cancer

The New NCD

There's no question that the new NCD - looking at both the draft and final versions - were intended to focus on NGS somatic (tumor) testing for CDx-type genes and drugs.   As mentioned above, the draft coverage section - which the public was able to comment on - focused on CDx genes and drugs, and required comparison to prior CDx tests, and required CED methods like RECIST tumor size measurement for CDx drug response.   There was no discussion of germline testing in the draft NCD, although the final NCD was edited to state that NCCN germline guidelines were, quote, "out of scope" and to state to the reader that "This NCD does not apply to all types of NGS testing." 

For a painfully detailed discussion of the policy issues, see my January 2, 2019 blog here, shortly after the MolDx LCD was revised.

CMS Covers Same Patient, Same Gene, Same Exon, Same Mutation by Sanger, not by NGS

The most unusual result of the MolDx LCD, as revised, is that it happily covers the same patient, same gene, same exon, same mutation, same purpose, same result, same report, if it is done by Sanger testing but not if it is done by NGS testing. 

As I've been saying since April 2018, this doesn't make sense based on Medicare's legal definition of when services are "reasonable and necessary."   Or, it just doesn't make any sense at all to oncologists, patient groups, lab groups, etc.   There's no price or coding difference.  CMS pays for BRCA 1-2 sequencing under one general code 81162 at the same price regardless of method.

The result also puts Medicare patients out of sync with the conservative but recent 2018 guidelines of the UK NHS genetic services for cancer risk (here).  They treat Sanger and NGS sequencing mutation reports similarly.

When Are You A Patient "With" Stage 1 Cancer?

The NCD is directed to tests "for patients with cancer."   Most women eligible for BRCA testing have stage 1 or 2 cancer, a cancer that is completely removed by lumpectomy or mastectomy.  So after that she no longer has demonstrable "cancer" - no imaging, signs, or symptoms of "cancer." 

So perhaps a post lumpectomy women is being managed for being status post Stage 1 lumpectomy, but she is not, in the literal sense, a patient "with a cancer" that you can point to.   In that case, the NCD doesn't apply to her at all. 

Another scenario would be a patient who had a lumpectomy 30 years ago - is she still a "patient with cancer" for the purpose of the NCD?  Could she have an NGS test for infectious disease, for example?   (Sepsis is a major cause of death in cancer patients, and commercial NGS tests are available, e.g. Karius; another NGS microbiology entrant is Ares.)  At what point does the NCD say such a patient is "a patient with cancer" or when does she stop being a patient "with" cancer?  What dose "with cancer" mean?

Do Small Cancers Block Unrelated NGS Tests Approved under LCDs?

MACs already cover some innovative molecular tests on NGS platforms such as the CareDx Allosure test, testing donor DNA as a sign of graft rejection, and the Veracyte Percepta test for work-up of suspicious lung lesions.

What if a patient eligible for either test also has (or has had) a small skin cancer, or in situ cervical cancer?  Are they disqualified from these two NGS-based tests?  If not, then CMS has stopped reading the NCD literally?  If yes, for how long are such patients "a patient with cancer"? 

Another twist is that when hereditary panels in Stage 1-2 patients are positive, about 5-10% of the time, the mutation may be confirmed with Sanger sequencing before reporting.  In this case, is what is being reported the Sanger confirmation?  Then it's outside the NCD, which only affects NGS reports.


CMS Might Say, NCD Text Drives This Interpretation

CMS might say, the most close reading of the NCD text drives this interpretation and leaves policymarkers no other choice.

I don't agree at all.

There is a body of law and policy going back well over a century to avoid interpretations of policy that don't make sense (to avoid absurdity).   This situation falls well within the scope of "avoid absurdity" doctrine, which I discussed in detail in an earlier blog.   One escape would be to read the internal text of the NCD ("Does not apply to all types of NGS tests" and "Daly et al out of scope") to modify the "header" section of the NCD.

This use of modifying text to implement unclear (or otherwise nonsensical) terms is normal in regulatory interpretation.  For example (discussed in same blog), the CMS regulation about Date of Service (42 CFR 414.500) has special rules for "molecular pathology tests" ...but doesn't define them.   Other CMS sources written later define this to mean only human gene tests, and not microbiology molecular pathology tests.  That is, the later CMS sources modify the words of the regulation.  Definitions inside NCDs can modify the header sections by the same token, especially when needed to avoid absurd outcomes.  We discover a cervical cancer NCD is about gynecologic cancers - not cancers in the "neck," the first dictionary meaning of "cervical" - by reading the body of the NCD or definitions therein.  It would be ridiculous to view such an NCD as applying to cancers in the neck because one was limited to reading only the title or header of the NCD and not the body.

For an Easter egg quirk in the 2008 LCD I mentioned, here.