Tuesday, January 28, 2020

Medicare's Revised NCD For Next Gen Sequencing in Cancer: Highlights and Quirks

On January 27, 2020, CMS released a revised NCD for next generation sequencing.
  • See the full CMS analysis here.
    • 41 pages as PDF; 33,000 words.
  • See my initial blog quoting the main decision points here.
  • In the cloud, see my lightly annotated extract of CMS's discussion of public comments, here.
In this blog, I'll recapitulate the highlights of the decision then describe some of the oddities.


Avoiding some details, let's summarize.  In February 2018, CMS issued an NCD guaranteeing coverage of FDA-approved NGS tests used on label in advanced cancer as companion diagnostics(e.g. the Foundation Medicine FDA-approved F1 test).   The NCD contains language implying that no NGS testing was covered (neither under LCD nor NCD) in non-advanced cancer.   CMS reiterated this stance in a transmittal in November 2018.  By early 2019, many stakeholders complained to CMS this would inadvertently block NGS testing of germline risk genes under guidelines (e.g. BRCA in high risk women post lumpectomy).   

In October 2019, CMS revised the NCD in a draft for public comment.  The revision included new sections on germline testing.   After public comments, CMS finalized the NCD on January 27, by inserting new language in and around the prior language, and by revising some of the prior language as well.     With the new language, germline hereditary risk testing is guaranteed national coverage in breast and ovarian cancer, if the NGS test is FDA approved or cleared (no genes are named).   For other cancers and for non-FDA tests, MACs can create coverage (or continue existing coverage).

In the new NCD, NGS tests for RNA (and oddly, "NGS tests for proteins" [sic]) are outside of scope.  Non-cancer tests, e.g. HIV genotyping tests by NGS methods, are also out of scope.


It's helpful to read the NCD analysis as a whole, and in particular you get insights into CMS thinking by reviewing the Q&A to public comments (here).

1. The Title

The revisions were put forward in October under the old title, Next Generation Sequencing (NGS) for Medicare Beneficiaries with Advanced Cancer.   The new final version shows the same title at top of page.  However, Appendix B of the final revision this week shows us the new title is actually changed to:  Next Generation Sequencing (NGS) for Patients with Somatic (Acquired) and Germline (Inherited) Cancer.

We don't usually categorize patients as "Somatic Acquired Cancer" and "Germline Inherited Cancer."  See also my point below #6.  

2. "Scope" Language New in Section A

Scope language was added to the top of the NCD, Section A.  

"This National Coverage Determination (NCD) is only applicable to diagnostic lab tests using NGS for somatic (acquired) and germline (inherited) cancer.  

Medicare Administrative Contractors (MACs) may determine coverage of diagnostic lab tests using NGS for RNA sequencing and protein analysis.  

MACs also have discretion to determine coverage of diagnostic lab tests using NGS for any non-cancer (e.g., infectious disease and heart disease) use.  These uses are outside the scope of this NCD."

The first sentence is a little quirky.  Tests using NGS aren't "for somatic cancer" or "for inherited cancer" but rather "for somatic mutations" or "for germline mutations."   

Test for for RNA expression levels by NGS (a technology which is increasingly usable)are excluded from scope because it's NGS for RNA and NGS for RNA is "out of scope," period.  Good, and clearly written.   For example, if the Mammaprint or Oncotype Dx tests are converted to NGS RNA expression platforms, they are out of scope.   

But I think such tests are also excluded since the scope is only for NGS based somatic acquired ^mutation tests, though the word somatic acquired [^mutation] is missing.  Still, the wording is a little odd.

2B.  NGS for Protein Analysis [sic]

It's  unusual to refer to "NCS for protein analysis" although at least this is "out of scope" so one need to worry about it further.

3. Single Test Use Language

The prior NCD had this restriction on repeat testing.  A test could be covered only if: 

[Prior Statement: patient] "either not been previously tested using the same NGS test for the same primary diagnosis of cancer, or repeat testing using the same NGS test only when a new primary cancer diagnosis is made by the treating physician."

The new language on repeat testing is found in both the germline and somatic mutation sections, and reads differently:

"[NEW: patient] not been previously tested [A] with the same test using NGS [and B] for the same cancer genetic content."

This is a little redundant.  If you are previously tested with literally the SAME test "A", then it was for the SAME genetic content "B," because the test was the SAME.   If the test is the SAME then the content of the test is the SAME.   Vice versa, if the genetic content of the test "B" is different, then the test "A" is not the SAME.    

I also wondered if this could be interpreted that you could use the SAME test but expecting DIFFERENT genetic content test results, for example, because the tumor has relapsed and therefore the tumor must have different genomic status now (different genomic content ^of the tumor), which we have to test for.  This fits the grammar and follows the legal rule that you should try to interpret a law so that no part is redundant and all parts matter (Scalia).   

4.  NCCN as a Guideline

The NCD analysis has two categories of guideline, "Evidence Based Guidelines" and "Professional Society Guidelines."  The draft in October placed the NCCN guideline under "Evidence Based Guideline" and then immediately criticized it for not being an evidence based guideline.  This was a little head-spinning.  They now acknowledge the NCCN guideline is "evidence based."

I think what happened is, there is indeed a kind of rigorous evidence based review that uses defined keywords, controlled literature searches, evidence admissibility criteria, meta-analyses, and so on.  NCCN would use that as an input, when available, but wouldn't produce it.  In addition, a comprehensive clinical guideline for physicians like an NCCN or ASCO guide has to consider all types of patients and situations, even those with less evidence.

5.  Lynch Syndrome "Not Enough Evidence"

CMS says it got submissions about Lynch Syndrome but there was not enough evidence.   Obviously, Lynch syndrome testing is included in all guidelines, covered by all MACs, by private insurers and so on.  So what gives?

I think the missing link here is that CMS was filtering only for Lynch syndrome evidence SPECIFICALLY conducted solely on NGS platforms.   That's not how the rest of the oncology and genetics community would sort the evidence for Lynch.

6. Weird Phrasing re Topics and Policy Categories

As noted earlier, the NCD refers to "acquired somatic cancer" and "germline hereditary cancer" rather than TESTS.

Weird things follow.   For example, look at new Section C, restrictions:

C. Nationally Non-Covered Indications
Somatic (Acquired) Cancer
Effective for services performed on or after March 16, 2018, NGS as a diagnostic laboratory test for patients with acquired (somatic) cancer are non-covered if the cancer patient does not meet the criteria noted in section B.1. above.

This suggests like a patient getting the FMI test who has acquired non hereditary breast or colon cancer is targeted under "C" non-coverage, but a patient getting the FMI test who has BRCA- or Lynch-driven breast or colon cancer is outside the scope of restriction section "C."  I would have thought CMS referred to somatic acquired tumor mutation tests but they specificallyemploy the words "patients with" not the words "tests for."