Wednesday, January 17, 2018

Very Brief Blog: A Final Visualization of the CMS NCD for NGS (and my PDF comment letter)

Medicare's wide-ranging draft NCD on the uses of NGS testing in oncology is posted for review here, and public comments are available for review here.

I offer a process chart that visualizes the NCD here:

A simpler view, summarizing just by patient types and test types, is here:

Several hundred comments will be on the CMS website, for which comments close on January 17.

I submitted a PDF public comment with these points:

The idea of catapulting Medicare oncology care into a learning healthcare system is terrific.   The specific approach, with sudden announcement of an NCD with complex and sometimes nonsensical rules, isn't the best way.   CMS should recall the Oncology Care Model was developed with many requests for information, town halls, public proposals, revisions, and stakeholder buy in.

Problems include four:

  1. Blocking any use of NGS in non advanced patients is a bad idea.  This blocks the whole world of germline testing (e.g. BRAC syndromes and Lynch syndromes).  It also prevents molecular tests (such as breast cancer prognostic tests) from being ported onto NGS platforms, since they are for non late stage patients.  
  2. LDTs have NO coverage except in NCI trials.   This is obviously a bad idea for many reasons, but a few concrete ones follow.   For example, FDA has freely approved cetuximab with only LDT KRAS tests, crizotinib with only LDT ROS1 tests, and Keytruda with only LDT MSI tests (the latter remains a current position of the FDA).   None of these drugs would have been available for on-label use for Medicare patients except in NCI trials, even after the on-label change at FDA.    CMS covered solid tissue panel tests on an LDT basis until a PMA test was available.  The same should occur in domains like germline testing (the whole world of ACMG and NSGC), leukemia/lymphoma panels, and liquid biopsy panels.   For germline testing, the best labeling some might ever get at FDA would be capped at 510(k) not PMA (not "high risk and life threatening devices").
  3. Some nonsensical positive coverage or perverse incentives are created.   For example, a hospital could biopsy on Monday, do five, $300 tests on Tuesday, and then CMS would additionally FMI F1 CDx for $3000 on Wednesday.   Any "recurrent" cancer is covered, without being advanced, which would include millions of small skin cancers (at least under CED).  There's no limit on paraffin block age.  Take home - if they got a million ten-year-old paraffin blocks from small recurrent skin cancers, they'd cut a check for $3B.
  4. Large numbers of patients are pushed into CED that makes no sense and, as written, is very invasive and costly.   If a CED project for $200M would be laughed out of the room as an NCI grant, it probably shouldn't be mandated expenditures under an NCD either.  Many patients will get hospice care, general chemotherapy, etc.   In one recent study (Tokaca 2018), only 13 of 66 lung cancer patients got a targeted therapy, but that is clearly what the CED envisions for all 66.   Patient reported outcomes won't change based on the brand of a test, any more than penicillin allergic reactions will change based on the brand of thermometer that diagnoses fever.  Simply creating a new system where genomic results are matched to drug records (Parts B & D, including time on treatment) would be a colossal improvement to what we have now.   Use best of breed stakeholder input and value-of-information design principals.  Avoid problems like patients going into multi year highly costly CED just because they are tested the day before their particular gene is converted to PMA level.  
I suggest limited scope of the NCD to solid cancers, for somatic gene testing, in tissue.  This leaves a grand and broad NCD in place.  90% of Medicare cancers are solid cancers, the NCD is solely and wholly designed for PMA gene-drug tests in somatic mutations, and 95% of the market or more is tissue biopsy.   In one swoop, for now, this avoids rolling avoid leukemia patients (or requiring absurd types of solid tumor CED), avoids a blockade of liquid biopsy coverage (which is working through FDA, commercial, NCCN, and LCD pathways), and germline testing.

There needs to be some safety valve for LDTs.   For example, the NCD could grant PMA test coverage (where they are available), grant CED coverage merely for the price of CED participation, but allow specific clear cut scenarios to be carved out of CED by having an LCD pathway.    MACs have been rigorous, tough, and slow to cover gene panel testing, so this is a narrow safety valve for narrow situations where CED would be unnecessary.   A two year moratorium on the non coverage paragraph would also be workable.

Simplify the CED.   Consider value of information strategies and allow rich public debate and comment on the minimum criteria set that has the biggest bang for the buck and burden.  Remember that some kinds of CED, like cardiac valve surgery CED, are primarily watchful: is the patient readmitted, does he have a stroke, must the valve be removed?   Oncology trial management is invasive and proactive, like monthly RECIST scans for PFS.   

Consider a CED pathway for liquid biopsy.   All solid tumors tests have a pathway to 510(k) status if they benchmark as NYS approved.   No such pathway yet for leukemia or liquid biopsy.  If CMS must, create additional entry points to the CED arm, one of which is NYS-510(k)-solid tissue, but another is NYS-liquid biopsy (since no 510(k) pathway exists).  The treatment is justified since liquid biopsy tests are in FDA pipelines, specific clinical uses are endorsed by NCCN, and some tests are NYS approved and are meeting commercial and MAC criteria for medical utility.


Added note (1/23/2018).   FDA has been markedly revamping its diagnostic test policies, including self-certification for some types of germline genetics, the NYSDOH-510K merged process for gene panel tests (MSK IMPACT), and the mere fact it now suddenly allows reporting of dozens or hundreds of "off label" genes in the Oncomine and FMI F1CDx tests.   Another trade wind:  On March 21-22, FDA will hold panels on whether it can downclassify tests like HIV diagnostic tests and HCV viral load tests from PMA to de novo 510K.  That's big.  Panel info here.