Thursday, November 30, 2017

CMS, FDA Announce Approval and Coverage of Foundation Medicine Test

On November 30, 2017, CMS and FDA jointly announced coverage of the Foundation Medicine Gene Panel Test through the Parallel Review process and the PMA approval.

The test is called Foundation One CDx or F1CDx.
  • The CMS press release is here.
    • Scott Gottlieb, head of FDA, assured Congress on December 7 that FDA is cooperating with CMS and this NCD will guarantee payment at CMS for all NGS cancer labs that cooperate with voluntary FDA PMA approval.
  • FMI press release here.  (Includes quotes from Lungevity and CTCA.)  
    • FMI press conference on the NCD, see the Genomeweb report, here.
    • FMI share price up 10-12% ($60>$65).
      • That brings share price back up to the 1Q2015 level.  In 2Q2016 it sank as low as $20.
      • Over the first week, FMI price has gone from $53 to $70 to $58 to $67.
    • 5 cancers or all cancers?  Since the CMS coverage includes all CDx tumors, and the FMI test includes MSI, and Keytruda and MSI are valid for all solid tumors, you might think the NCD would cover FMI in all solid tumors via the CDx MSI.   
      • Apparently not.  
      • Apparently MSI is reportable but doesn't have CDX status yet, because FMI is only claiming new patients under the NCD for a few cancers like lung, not for all cancers via the MSI angle.
  • Coverage at MedCity News here.  Xconomy here.  Genetic Engineering News here.  Fierce Biotech here.  Genomeweb here.  Blog by CureOne [Med-C] Dane Dickson MD, here.  Coverage at New York Post, here.
  • The CMS NCD is on the web here 
    • PDF hereTrack public comments here.
    • This is a "draft" NCD, with 30 days public comment through December 29.
    • At 46,000 words and some 62 pages (as PDF), it's one of the longest NCDs ever.
  • The FDA press release is here.
  • FMI FDA PMA Review Documents...Missing in Action.
    • The Foundation One PMA Webpage is here, but as of December 8, only the summary data is posted, none of the FDA review PDF documents.
    • The PMA number is P170019, the Class III PMA Product Category is PQP.
    • Summary of FMI versus Oncomine labeling, here.
  • Weird fun fact.  The 11-page NCD request (from FMI) is dated November 17, 2017 (!).
  • Scope and cost.  The FMI NCD request estimates a potential annual volume of 46,000 cases (at 20% penetration into the relevant CMS advanced tumor population of 230,000 incident cases.)  
    • At $3000 per test, that's about $140M dollars per year.  
    • But if you included all surviving patients backwards a few years, and for example all women with a 1 cm breast cancer and 1 positive lymph node in the last 5 years, the patient population would be much bigger.
  • This NCD provides another reason to fix the goofy update to the 14 day rule that appeared recently.  
    • The 14-day rule was liberalized to provide better CDx access to Medicare cancer patients, but "excluding genomic sequencing procedure" tests [!], which the F1CDx test is.   More about 14 day rule brouhaha here.
  • Huge dive for MSK 510(k) IMPACT test.
    • The MSK IMPACT test was FDA de novo cleared by FDA last month with huge fanfare.  However, the NCD gives much lower status to this clearance route, covering it only in onerous CED trials.  (For example, do a RECIST study monthly at $1000/month for one year and you've spent $12,000.)
    • I think CMS wanted to distinguish somehow between PMA and 510(k) gene panel tests, but the NCD proposes to make 510(k) tests use impossibly costly registries.  Of course, it's not just the labs, it's all the cancer patients who don't have a cancer specific PMA gene and NGS test available.
The FDA writes that with 324 genes and "two signatures," the F1CDx  test can identify which patients under 5 tumor types may benefit from any of 15 FDA targeted treatments.  The two "signatures" are for MSI and for total mutational burden.  Thus, the test creates an FDA-approved route to a TMB assay result.

CMS coverage is for patients "with advanced cancer, e.g. metastatic," with no prior NGS test, who "continue to seek further cancer therapy."  "Coverage" occurs when there is at least one on-label FDA approved diagnostic and at least one CDx drug in that kind of cancer.   "Coverage with Evidence Development" occurs in other situations, which are complex but generally distinguish one track of CED for FDA approved/cleared tests used in registries, and and another CED track for LDT tests used in NCI trials.

Synchronized Harold Varmus Op Ed in Science

In parallel, coordinated but subscription-only editorial by Harold Varmus appeared in Science today, arguing that payers should require CED in genomic test coverage, as well as use of genetic data registries.   See Science here, see Genomeweb here.

Much more detail after the break.

CMS Deep Dive

There are criteria for "patient" and for "test."

The "patient" must have advanced (e.g. recurrent or metastatic or Stage IV) cancer, not prior testing with the same NGS test, and seeking further treatment. 

Then it gets tricky.   I would present it this way.   For the "test," there are three different pathways, one coverage pathway Pathway 1 for FDA approved on label tests, and two CED pathways (2a) and (2b) for FDA approved/cleared and LDT tests, in either case, used in registries or trials. 

click to enlarge

True Coverage: Pathway 1

Under coverage pathway "1," the test uses NGS and is FDA approved, the patient has "a cancer with an FDA approved CDx indication," and the test report is FDA approved and specifies FDA indicated treatment options.

CED Coverage: Pathways (2a) and (2b)

Under CED pathway (2a), the test must be FDA cleared or approved, the laboratory must be in the NIH Genetic Testing Registry, the patient enrolled in a prospective registry, and the registry addresses how outcomes compare to "initial validation of the CDx" or "a cohort of controls receiving the same treatment" [presumably unstratified?], or studying how clinical characteristics affect the registry patients versus those in "initial clinical studies."  All the following shall be tracked:  overall survival, PFS, objective response by RECIST, and "patient reported outcomes."

Under CED pathway (2b), The test is a "report of test results used in the management of the patient's cancer."   (No reference here to FDA approval.)  But: the test is supplied within an NCI clinical trial.  This NCI clinical trial must adhere to various additional criteria, but it probably would.

Noncoverage is proposed for NGS testing outside the above criteria.

This was a summary.  For full details, see the actual original NCD at link at top.

The NCD also provides the public with 5 questions that CMS particularly hopes the public will address.  Here.

Basket NCD - Basket of Different Tests

The NCD notes that while it was initiated under request by F1CDx, there have been "three other genomic oncology panel tests for advanced cancers" and the NCD is for "NGS oncology panel tests" rather than only F1CDx as a brand specific NCD.

The NCD calls out the following tests:
  • FMI BRCA CDx test, 
  • the F1CDx, 
  • the Oncomine Dx Target Test, 
  • the Illumina Praxis RAS Panel, and 
  • the very recently cleared MSKCC IMPACT test.

Comment is open for 30 days until December 29. 

1.  True coverage is only for PMA tests.
True "coverage" seems to be open to FDA approved tests.
"Merely" FDA cleared tests like MSKCC IMPACT are relegated to CED 2-a (registries).  This is also where you find PMA tests used in cancers with zero CDx drug options.   LDTs are relegated to CED 2-b (NCI trials). 

2.  NCD covers any NGS based diagnostic used in a solid tumor.
The NCD does not appear to refer directly to liquid biopsy versus paraffin NGS tests.  Rather, it applies to solid tumor cancers where NGS method testing is used.

The NCD is so focused on "NGS" (a broad term) that it might distinguish between some test types that are "true NGS" or "not true NGS." For example, currently clinically used classes of LBx tests have different underlying methods such as NGS for one and digital droplet PCR for the next.  Also in paraffin tests, large comprehensive hotspot tests are not controlled by the NCD as long as they are not "NGS."  On the other hand, if you did four tests (EGFR, KRAS, ALK, BRAF) by NGS, it would fall under the NCD and be covered only in NCI trials if it is by LDT NGS methods.   This means a 95% accurate PCR EGFR test would be outside the NCD and covered, but a 99% accurate NGS EFGR test is covered only in NCI trials. 

3.  Sudden drop in status for MSKCC IMPACT test?
Just days ago, there was huge policy interest in the New York State, de novo 510(k) track for NGS panel tests.   However, the NCD  covers 510(k) NGS tests only in elaborate, costly long term high quality registry research trials, which would likely cost far more than the test pays.  This would in practice put a hospital lab cost center far into the red, assuming a long term debt for running a trial with monthly RECIST imaging tests and tracking survival as the patient moves among future locations and institutions.

4.  Huge differences among currently PMA approved tests.
While the NCD refers to all PMA approved tests, as of today, there are huge differences among them.   FMI F1 is by far the broadest, and includes MSI and TMB.   Thermo Fisher Oncomine is next, but the current FDA approval language permits reporting of 23 of the 70 genes on the full Oncomine NGS panel, and the labeling is only for lung cancer right now.   After that, the Illumina RAS test is a "foot in the door" test for that platform, and the FMI BRCA NGS test similarly was a foot in the door test. 

Currently, there are no FDA approved or cleared large gene panel NGS tests in liquid biopsy, an important emerging clinical area.

4.  Just who can afford to run these CED registries?
One issue is that NGS panel test coverage for patients off label from FDA approved tests might appear to be only open to patients getting tests tests from very large "over funded" labs and large "richly funded" medical centers, because of the high cost of the required registries which will greatly overflow CMS test payments,  and the requirement to track all tested patients all the way to overall survival, which could be many years.  All these burdens are absent for on-label FDA NGS tests.   It's difficult to know how CMS can enforce the long duration CED; for example, if you pay for a test today, how can you prove the lab did or didn't follow survival for 5 or 10 years for the mandatory OS rule.

5.  Navigating the Implications of the Text
For coverage, the NCD seems to require that an FDA approved CDx exist, for the patient's cancer. OK, that's going to be a one standard list of cancers at any point in time.  If someone comes up with a CDx drug for squamous skin cancer, then that cancer will be on the list.

On the other hand, tumor of unknown origin, for which gene panel tests might be very useful,  is left out?  Or maybe the TMB or MSI CDx applies for immunooncology (based on MSI marker) in TUO, so TUO is in.

5b.  What about Keytruda and the List of Cancers?
As of this year, the labeling for Keytruda (pembrolizumab) covers "the treatment of adult and pediatric patients with unresectable or metastatic solid tumors that have been identified as having a biomarker referred to as microsatellite instability-high (MSI-H)."  If so, every cancer (gall bladder cancer, appendix cancer, adrenal gland cancer...) needs MSI testing for Keytruda.  If so, every solid tumor (advanced) cancer patient would be a candidate for testing under Type 1 Coverage in the NCD.

6.  Test Reporting Governed by NCD, not Doctor's Therapy Choice
The NCD also requires that the report indicate FDA approved drugs.  But it doesn't extend to guiding or controlling the drugs the patient actually will get.  So if you have advanced lung cancer, that's a cancer that qualifies.  If the report lists the FDA approved drugs in lung cancer, the report qualifies.  Then, it's left to the judgement of the physician to pick an on or off label drug and move forward.

7.  World's most expensive MSI test.
It sounds like if the patient has already had standard testing for all the major lung cancer genes, or even a prior F1 LDT or at least a similar test last year, he can now get an F1 CDX test just to get TMB, if the doctor orders it, or just to get an MSI result.  It seems to fit the rules and the letter of the NCD.

8.  New Genes and Delay for PMA.
Sometimes, FDA has approved drugs before the PMA test (such as BRAF in lung cancer), or the gene test is well known and widely employed before a PMA test exists (as happened with KRAS).  The NCD doesn't allow for this.  Such a test would  be an LDT and trigger the NCI only pathway until if and when an FDA diagnostic is finalized.   

9.  Market Size.
In the FMI request letter, they state there are 54M Medicare beneficiaries, who get 895,000 new solid cancers per year, and the progression (or incidence) of advanced cancers is 230,000 per year.  They list current gene panel test penetration at 10% (23,000) potentially rising to 20% (46,000).   At $3000 per test, that's about $120M per year.  But there's a strong incentive to test.  Pubicly available FMI announcements give the fully loaded cost of testing at $800, and the CMS price for 81455 is about $3000.   Therefore, there would be alot of incentive to test the 80% or 90% of solid tumors not currently tested.   Testing all 230,000 incident advanced cancers would max out around $800M in 2018 and if you added 100,000 surviving patients who got advanced cancer in 2017 it would be another $400M market.  That gives a total addressable market of $1.2B in 2018.

10.  No comment on old specimens.
The NCD makes no comment on old archived specimens.  If you had breast cancer in 2010 and it is recurrent in 2015 or 2018, sounds like the NGS test on the archived 2010 paraffin block would be covered.

11.  Analytical accuracy versus initial PMA validation of a gene-drug pairing.
The NCD makes much of the historical PMA validation for all CDx and the use of merely analytically validated test (left far lower on the totem pole).   However, the FDA and everyone else realize that analytical validation and accuracy drives the outcomes of gene-patient-drug combinations.  The NCD may seem increasingly dated in pretending that for every gene (among hundreds of genes) and thousands or millions of variants that is some differencebetween an analytical true result and PMA validated's a viewpoint that comes close to "magical thinking." 

12.  Coding Implications?
CMS tends to like to have clear cut coding that matches NCD decisions.  CPT codes make no distinction between NGS and hotspot tests, so CMS will likely either make complex new classes of modifiers or make its own code set for NGS testing in tumors.   (For example, it might require 3 sets of codes for Type 1 coverage and Type 2a 2b CED.)  Or it might require that each test be coded for what class of coverage it fits into.  And the tests now covered seem pretty diverse; would there be test specific pricing?   Would CMS require PLA codes or ADLT codes?   To be continued.  If CMS really wants to implement the NCD in 30, 45, 60 days, it would have to decide quickly.

13.  Timeline?
There is only one other case of Parallel Review, where a draft NCD appears suddenly and a final NCD after 30 days public comment.   This is the  Exact Sciences NCD, where public comment closed on September 10, 2014 and the NCD was issued on October 9, 2014.  That was an uncontroversial decision and I believe had no appreciable change between draft and final.   This NGS NCD will be much more heavily reviewed and could likely take longer to finalize. 

14.  Odd Use of Data for PMA Only Coverage And Exclude LDTs Non Coverage
The NCD reviews the literature on using broad panel tests in solid tumors, rather than only 2-3 targeted tests.   Generally, such literature reviews the whole panels (e.g. 300 genes) and not the incremental benefit after the first 5-15 genes.   Regardless, all the world clinical literature used to justify broad gene panel testing is based on LDT NGS tests, yet the conclusion is that only PMA NGS tests are fully covered.

15.  FMI F1 CDx May "Include" MSI and TMB Reporting, but Not at PMA Level?
FMI seems to say that it gets coverage in 5 cancers, but the FMI F1 CDx test includes two signatures, for MSI and TMB, which might apply to all cancers.  This might mean that those signatures are included in the test, but not endorsed by FDA at PMA level.   As of December 3, I haven't seen the PMA paperwork online at FDA.

16.  Use of CDx as a Simple Pivot Point for NGS Panel Coverage
Read carefully, the NCD uses only the several CDx diagnostics as the "pivot point" for a decision for test coverage.   Yes, if you use a Roche single EGFR test or a Foundation Medicine Panel Test including EGFR, either way, there will be just the same improved patient outcomes when that Medicare patient gets his EGFR drug.   And CMS will pay whether you did the Roche EGFR test for $300 or the FMI test for $3000.

The NCD doesn't really address how much benefit is obtained from the "extra" 300 genes.   (There may be benefit; but it doesn't have to be proved rigorously by CMS, since the NCD pivots on the one gene.)   On the other hand, there isn't anything in the literature cited in the NCD that an FDA approved CDx that reports a certain EGFR mutation leads to better results than a non PMA test finding just the same mutation (and to imagine there is a difference is probably close to magical thinking).

The original LCD from Palmetto MolDX covering Foundation One in lung cancer two years ago also took a quite narrow logical approach to benefit, arguing that since a few patients per 100 negative for hotspot EGFR tests would be found with Foundation One EGFR sequencing, therefore, the Foundation One test was reasonable and necessary.   (That is, the Palmetto LCD also didn't need to address how much benefit the extra 300 genes conveyed.)

17.  Would this NCD shape the future market?
Potentially, if finalized as-is, the NCD would substantially shape the future market.  (Put on your George Soros hat here).   Coverage for any NGS test, no matter how small or large, is predicted on having one pivotal CDx gene in each major cancer.  The smartest test would have 1 CDx PMA gene in lung, prostate, colon, and breast cancer (the largest cancers), and overall, be as large and expensive as possible (e.g. 1200 genes for $8000?).   Coverage of the test would be guaranteed in all those patients at the ADLT price, so the company would want to extremely closely control sales via private insurers.  Development risk would be very low as long as you could accomplish the minimal needed bridging studies for each PMA, and reimbursement risk (for Medicare) would

18.  Would FMI benefit from becoming an ADLT?
Probably not right now.  It seems to be widely remarked in the financial press that their average private payer reimbusrement has been around $2600.   So they are better off at the standard plain jane PAMA 81455 price of $2900.   If their own market price ever rises much above that, they could convert to an ADLT of the FDA approved sole source type.

The NCD coverage text, the "operational" part of the NCD, is also provided here.

The FMI BRCA NGS test approval page (2016) at FDA is here.  PMA Product Class PQP (NGS panels) is here.  A December 4 "approval statement" and label is online, but I haven't seen full PDF PMA documentation yet.