Sunday, October 30, 2022

Journal Club: Interesting Titles from CAP TODAY, from JACR

A flurry of interesting journal articles.


From the October CAP TODAY, two cover articles.  One on "the art and science of positive blood cultures," which may sound like a yawner, but it's about the range of new molecular and fluidic biotechnologies that are changing the field of clinical  microbiology, especially in the area of rapid upfront results.    See also "checklists now made to fit for NGS labs," about updates and new paradigms for CAP policies and reviews for NGS.   Find it here.


I've previously remarked on how much cool stuff appears in JACR, Journal of the American College of Radiology.   From the October issue:

  • Vachani et al.  Complications after biopsy of pulmonary nodules.   Complications rate 25%, dominantly pneumothorax, of which, 32% (8% of total) require a chest tube.  Here. Open access.
    • Pair with an in press article on costs of lung biopsy, Tailor et al., here.
    • These are relevant to all the developers of tests to biologically predict the risk of cancer in a small ambiguous imaging nodule.  One test of this type, Biodesix CL2, 0080U, had 1700 Medicare uses in 2021, for $6M in payments (here).
  • Hendrix et al.  Radiologist preferences for AI-based decision support during screening mammographyHere.  Open access.  
    • This is a very original, very detailed article on different preferences and needs of radiologists for AI in screening imaging.  See esp. Table 1 for their approach.   
    • Lab medicine: My point here is, this approach could be a model for thinking about topics like digital pathology AI, and even for physician thinking, use and interpretation of NGS reports (aside from AI).  
  • Makhnevich et al.  Comparison of chest radiograph impressions for diagnosing pneumonia: Categories of language uncertaintyHere
    • We have uncertainty in anatomic pathology and cytology, as well as molecular medicine and genetics (consider also VUS).   
    • This is the most elaborate and structured paper I have seen about levels of uncertainty.
  • Ho et al.  Prioritizing screening mammograms for immediate interpretation and diagnostic evaluation, based on : risk of recall.  (Meaning, risk of calling patient in for more studies).  Here. Open access.
    • Obviously the potential for a notable contribution to patient/physician service and public health.
  • Adleberg et al.  Predicting patient demographics from chest radiographs with deep learning.  Here.   
Selected papers in each issue of JACR also get a "visual abstract" (from the publisher, in similar format) to aid the reader quickly absorb the flow of the material.

"Standard" visual abstract designed by journal for selected pub's (JACR)

Neogenomics Press Release: Insight into MolDx and Minimal Residual Disease (MRD)

 On October 28, Neogenomics, a publicly held lab, issued a press release on the status of its efforts to get "MRD" (minimal residual disease) approval from MolDx.   Find the press release here:

Note that based on an earlier press release, this decision had been under review at MolDx from Jan 2022 to Oct 2022.

What the Press Release Says

"MolDX informed the Company that additional clinical evidence is needed in order to secure Medicare coverage for RaDaRTM for Colorectal Cancer (CRC). Specifically, MolDX asked for a direct comparison with other MRD tests being utilized or a full clinical study."

My understanding is that Neogenomics' MRD assets come from its Inivata acquisition; the Neogenomics website links through to the Inivata one.  The press release also remarks that Neogenomics would pivot from CRC MRD to breast cancer MRD.   (See a follow up article in 12/2022 on Neogenomics' progress with breast cancer MRD.)

On October 28, Neogenomics stock was down a little (6%) or about $60M on a market cap of $1B.  Natera was up 2% or about $90M on its market cap of $4.6B.  

What the LCD Says

Find LCD L38779 here (see also, however, billing article attachments to the LCD, AND tech assessment spreadsheets, the latter on the MolDx website).  When the LCD was released, almost exactly a year ago, I referred to its publication as "a dramatic day for precision oncology" (here).   

The LCD provides the potential for coverage of MRD tests for many different conditions, from different labs, within a four-box framework.  The two axes are "solid organ cancers" vs hematopoietic ones, and "MRD for relapse/recurrence" vs "MRD for treatment response determination."  However, any particular test, lab, and coverage indication is designed to be enumerated, one by one, on a rolling basis.  

I think the Neogenomics press release is consistent with my reading of the LCD (and my hearing multiple talks by MolDx authors).   MRD tests are all different, with their own approaches to exome testing (or the creation of tumor-naive tests as well), and circulating tumor DNA (ctDNA) may vary between types of cancers.   Therefore, MolDx takes the position it's not enough to just (for example) say that your test is negative in these 20 cancer-free patients, and positive in these 20 with known metastatic cancer.   Rather, you have to show the test actually picks up cancers at the new MRD stage, and accurately, in patients with that cancer type and earlier than imaging (or another standard of care.)   (See my footnote on clinical trials. FN1).

Per the press release, MolDx is also communicating a bridging may be acceptable; for example, the covered Natera CRC MRD test is positive in these 25 patients and negative in these 25 patients and your test reports the same thing in each patient.   


MolDx has approved only a relative handful of tests and cancers: The Natera MRD test for recurrent CRC or bladder cancer; the Natera treatment response test for checkpoint inhibitor therapy assessment; the Guardant test for CRC recurrence.   The barrier to entry for getting a new lab, a new test, into the LCD MRD system, remains pretty high.



I don't have any special knowledge on the scale or design of a trial acceptable to MolDx.  However, here's how I think about it.  Consider screening tests for colorectal cancer at FDA.  Typically, you need circa 100 positive colonoscopy patients to get tight statistics on your new test.  But at a 1% positivity rate, you need to screen about 10,000 negative colonoscopy patients to sift out those position 100 ones.  So CRC prevention tests for FDA tend to enroll about 10,000 patients.   

Turn to MRD.   Let's say hypothetically that colorectal patients have a 10% relapse rate at one year.  Let's say you need at least 20 to have statistics on MRD performance (all 20 relapsed on imaging at time T(image), and each will have a detection of relapse with ctDNA at time T(mrd).)   You need data that, for example, your test picked up relapse before imaging 17 times, the same as imaging 2 times, and the mrd test didn't ever turn positive 1 time (N=20).  That's just a for-example.   To get those 20 relapses patients, you'd need 200 enrollees, each followed for a year, so the trial duration is from the first enrollment day to the last enrollment day plus 1 year.   Again, this isn't any promise of what MolDx wants to see, but just gives the reader a beginning way to begin to think about enrollment in an MRD test trial.   

Endnote: Volatility in the MRD market

On October 28, Neogenomics stock was down a little (6%) or about $60M on a market cap of $1B.  Natera was up 2% or about $90M on its market cap of $4.6B.  (Guardant was unchanged.)

Remarkable are the 1-year share price ranges in this industry.   Natera's one-year price range is $27-121 (current, $48).   Neogenomics' is $6-47 (current $7.50).  

Saturday, October 29, 2022

Kaboom: CMS Releases 2021 Spending by CPT Code; Shocking Molecular Test Abuse Rises

For several years, since 2018, there has been a shocking growth in the use of  Tier 2 codes in Medicare Part B, with a significant portion of labs using these codes in high amounts being indicted by Department of Justice.  

  • See big Department of Justice press releases around September 2019 as Operation Double Helix (here) and again in August 2022 (entry point here).  

You can just use the "fingerprint" pattern of super high Tier 2 code billing (by indicted labs) as a "magnet" to look for similar lab billing patterns.   

The billing patterns are lab operations dominated by all the Tier 2 codes, and often only the Tier 2 codes.  These payment patterns, which are over the top unbelievable, are almost entirely in Novitas and FCSO MAC regions.  Since it is  impossible imagine that one Medicare patient needs all of the Tier 2 codes on one day, let alone 1000 patients in a row, I call these "Highly Unlikely Codes." 

On a smaller scale (topping out at $200M rather than $280M) is labs dominantly billing just for 87798, "other pathogen, PCR."  At least through CY2020, almost all 87798 billing in large amounts occured in MolDx states. I also discuss this below, and include bar charts by year.

CMS has now released CY2021 data, on October 28, 2022.   Total lab industry spending (including Anatomic Path) was about $8B before the main COVID codes, which added just about $1B more.   Of this spending, $2B was in anatomic pathology.

Find 2021 data here:

Molecular Pathology Spending

By my best (though rapid) estimates, MoPath codes, including PLA codes and molecular microbiology, tallies about $2.8B before the main COVID codes.  Add a billion for the main COVID codes, and mo path rises to about $3.8B.   

For the rest of this analysis, I will leave off the main COVID codes, and work from the $2.8B number.

Unlisted Code is Biggest Code

The Unlisted MoPath code, 81479, rose to $409M, up from $290M the year before (up 40%).  

At least in all past years, 95-98% of 81479 billing is in MolDx states.   

  • For example, Myriad MyRisk testing falls here under 81479, per MolDx coding instructions to Myriad. 
  • Increasing use of the 81479 code means that more and more of the Medicare molecular payment apparatus works separately from the statutory PAMA rate setting process.  
  • Increasing use of the 81479 code means it is impossible for outside researchers to know what genomic services Medicare patients are getting, making genomics unlike any other healthcare service (like "appendectomy" or "cardiac ultrasound" or "flu shot.")  See next paragraph.

Comprehensive Genomic Profiling (CGP)

Public health experts interested in precision medicine would probably look to growth in the comprehensive genomic profiling codes, 81445, 81450, 81455.   Use of these codes remained small.  81445, 5-50 solid tumor genes, with 2,505 uses at $1.5M (about $600 each).  81450, 5-50 hematopoietic cancer genes, with 16,040 uses at $12M (similar price).   Finally, 81455, 51+ tumor genes, any cancer, with 4,310 uses at $12.4M (about $2900 each).   The  4,310 uses of 81455 are dwarfed by, for example, 21,223 uses of Foundation Medicine 0037U, $74M, and likely similar usage of Caris comprehensive testing billed as 81479 under MolDx.  In short, a health researcher looking only to 81445, 81450, 81455 which are the regular CPT codes, would see little of the actual use of CGP.

Proprietary Codes

Leading proprietary codes are Cologuard at $253M (up 20%), Oncotype at $93M (up 20%), Foundation Medicine FDA (FFPE) $74M (down slight 4%).  These 3 proprietary tests are 15% of all MoPath spending.  

Highly Unlikely Codes (red)

Before CY2018, when they were first assigned prices, Tier 2 codes were almost unknown in Medicare Part B billing.  

Dollar volume shot up in 2018, 2019, 2020, even as DOJ was issuing press releasing that it was 'catching' bad actor labs, which, in data not mentioned by DOJ but discussed by me, the billing profile was Tier 2 billing, over and over again.  And only in Novitas and FCSO states.   Very, very, very little of spending for Tier 2 codes are labs like Ambry, GeneDx, Quest, etc.   They can be labs with names like "ABC Lab" which by the time you hear of them, Google to be a small abandoned storefront in a mini-mall by a tattoo parlor. 

Because of the years of attention, I was on the edge of my seat to see 2021 data.  IT IS BAD.   Tier 2 Level 9 rose  from $207M in 2020 to $283M in 2021.  That is, it got WORSE in the FOURTH year.  Much worse.   Adding often-dubious code 87798 (pathogen, other, PCR), about 30% of MoPath billing was at risk spending, to say the least.  

This seriously distorts any analysis of overall growth (or medical needs  imputed by usage profiles) for mopath spending in Medicare Part B.

See my videos about the scandal (overview here) and (one lab analyzed here).  Press releases with CMS quotes talk about the dedication and tireless brilliance of the fraud detection efforts, but the stuff being done (by looking up billing records) is grossly obvious, can be shown in seconds in Excel, and you could train a 10 year old to recognize it in five minutes.  And it's just the same in 2018, 2019, 2020, and 2021 - there has been no evolution of the abusive and unbelievable billing patterns.  And bizarrely, CMS pays for these Tier 2 genes in multiples as high as 5 (!!!), which is mind-blowing, if your mind isn't already blown to the ceiling.

Update: By November 2022, Novitas had put in belated billing article defenses against rampant Tier 2 code abuse - here.


The top 5 codes had 44% of all spending, and 20%  (about half) was gravely skeptical codes 81408, 87798.  This is just plain crazy.   

If some burocrat at CMS had just taken 5 god-dam minutes to change the MUE edit of 81408 from "2" to "1" (the NORMAL MUE number for genetics codes), it would have saved nearly $150M.   (Assuming that took 5 minutes, it would amortize to saving about a billion dollars per hour of admin time.)

The top 20 codes had 72% of all mopath spending.  The top 10 codes had 57%.  Only 2 of the top codes were "PLA" codes, and both were FDA approved (0037U FMI FFPE, 0242U, G360 FDA).

PLA Codes

Of $2.8B in MoPath spending, only $230M went through PLA codes (or MAAA U codes).  Of this, 85% went through the top 10 codes, and over 50% went through the top 2 codes (being Foundation FDA PPFE, and Guardant FDA G360).  Only 75 of some 300 PLA codes were paid even $1000 by Part B. See the second chart below.

Screen Shots - Click to Enlarge

Top 20 Mopath Codes, 2021.  (Red = Atypical, Green = Proprietary)

PLA only:

My data as cloud spreadsheet here.

Spending for Tier 2 codes, shown below, in bar chart from 2017 (baseline) through 2021.  Total spending on 81408 was $283M, being almost identical in 2019 (year of "Operation Double Helix") and in 2021 (last year).   (!!!!)

I've subtracted the baseline year, 2017, from 2018-2021 spending to get the 4 year excess spending on tier 2 of $857M.  See excess spending bar chart below.

Almost no spending for 81408 went through NGS MAC or MolDx MACs; spending via Novitas MAC and FCSO MAC.  Most labs billing 81408 billed it in multiples of 2 (occasionally, 4) - that's two full sequence of what the AMA defines as very rarely needed genes, per Medicare patient, one patient after the next.  

Spending on 87798, Other pathogen, PCR.  I've used 87799 (Other pathogen, quant PCR) as a control (see bar chart below).  

At least through CY2020, labs billing 87798 were overwhelmingly in the MolDx states, with little usage in Novitas, FCSO, or NGS MACs.  MolDx has since rewritten its pathogen  reimbursement policy.  Spending on 87798 in 2018-2021 was $537M, and if we subtract 2017 ($25M) as a baseline, excess spending on 87798 in 2018-2021 was $437M.

87798 2016-2021; 87799 as control

87798 (for CY2020) by State (MolDx red)

Are you keeping track?  
If excess spending for Tier 2 codes in 2018-2021 (over 2017) is $857M 
and excess spending of 87798 in 2018-2021 (over 2017) is $437M, 
that totals $1.3B.

The Genomic Health (Exact Sciences) prostate test (#5 on the PLA subsection at $16M) has since been sold to MDX Health, here.   The three major prostate prognostic panels for Cy2021 are Decipher at $37M (now Veracyte), Myriad at $29M, and Genomic (now MDx) at $16M.  (I think the year before they were closer together).

I'm more than dubious of Noonan Syndrome 81442 being one of the highest-billed codes in Medicare genetics.  This made no sense to genetic counselors I talked to, and volume on this likely-forgotten and unedited code rose from $0 in 2019, to  $8M in 2020 (from a small number of little-known Southern labs) to $35M in 2021.  In 2020, half the volume was from one lab you've never heard of (not from Ambry's, or Quest's, etc).   If a code skyrockets in use, and the code has no place in the Medicare population, and the code is billed only by labs with names like "ABC Lab" in Florida and Louisiana, it's a bad sign.  

I think that cardiology genetics is potentially very important in the Medicare population, but knowing the status of coverage policies in CY2021, I'm not sure if the high billings of aortopathy panel, dup/del analysis, 81411, was from fully legit sources.  Billing was $0 in 2019, $9M in 2020, and $48M in 2021, again from little-known labs and in a very few of the 50 states (FL, LA, TX).  

That's right; in 2019/2020, Medicare paid out $100M for 81442/81411,  which were paid $0 in 2019.  Pretty big change in...standard of care.  In Florida and Louisiana, at least.

COVID spending.  Nearly all CMS COVID spending in 2021 went through high through codes U0003, U0004 ($75 each) and U0005 ($25  48 hour bonus).  U0003 predominated over U0004 (8.7M vs 2.2M cases.)  There were 10.9M cases of U0003/U0004 together, and most of them (9.2M, 84%) billed along with U0005 and got the $25 bonus.  Dollars were U0003/4 at $818M and U0005 at $229M, or $1.05B.

Prostate MAAA tests.  Prostate MAAA tests are led by Decipher 81542, Myriad 01541, Oncotype 0047U.   These three collectively grew 47%, with group sales of $82M, but pulled farther apart from one another in 2021.  Decipher grew 89% to $37M, Myriad 37% to $29M, and Oncotype only 5% to $16M.  

Demographics.  There's a database where you can get 2020 demographics by lab, I've listed demographics for Foundation Medicine as an example, here.

Demographics of highly suspicious genetics labs.  I checked one against a CMS provider-level demographic database (same as for Foundation in prior bullet).  Genetic testing of pediatric genes for a population 14% over age 85, (!), 30% with Alzheimer's (!), getting exactly 12, individually ordered identical genetic tests each and every Medicare patient, and for diverse  rare or pediatric diseases.   Yeah, it would be hard for CMS detectives with years of experience, supercomputers and AI to find a hint of anything fishy here.  (They were also 38% dual benes (poor), and 14% Black (racially balanced).)

Corrections made.  Due to a sorting error, 81162 BRCA was omitted from the first edition, but corrected around 12N ET on 10/31.  It should be the 5th highest code, and raises the total from $2.7B to $2.8B.

Friday, October 28, 2022

Very Brief Blog: Trends and Concentration in Healthcare Lobbying

Schpero et al. in JAMA Health Forum report on trends in healthcare lobbying, and concentration.

  • Health care lobbying dollars have been surprisingly steady 2010-2020.   
  • Spending is highly concentrated, with the top 10% of providers or biopharma, putting in 60-70% of the dollar volume.

New LCDs in Urine Drug Testing; Florida Cans LCD for Immunohistochemistry

CMS posts new draft LCDs on Thursdays, as well as new finalized LCDs.

Urine Drug LCDs

I don't work on urine drug testing myself, but I know it's a big industry.  Palmetto posts a revised LCD here, and FCSO (Florida) a revised LCD here.   They're not the same, but they're long and I haven't analyzed how different they are.  And I would think each MAC would have high prior urine drug LCDs, but they're treated as new ones, not revisions.

Immunohistochemistry LCDs

Meanwhile, FCSO posts for deletion (retirement), an LCD for "histochemical stains and immunohistochemistry."  L36234, here.  The 'revision explanation" text says, This LCD is being retired secondary to data analysis and a review of the current literature, which revealed that the LCD and its corresponding Billing and Coding Article are neither reflective of the current standard of practice nor having a significant impact on Medicare expenditures.

Note that Palmetto has an LCD wit the identical title, not posted for retirement (L35922).  There was a big brouhaha when it was introduced in October 2015.  

Medicare Nerd Notes x 2

Until recently, MACs didn't need to post advance notice of an LCD retirement, they could just pull it down rapidly.  Now, they post an LCD, and you have to look to the "reason for revision" box down at the bottom to see the only reason for posting, is retirement.

The Palmetto "Immunohistochemistry" LCD was originally a "MolDx" LCD (so titled).  The word MolDx was deleted from this LCD because now MolDx only handles DNA RNA tests.   (Both human and pathogen).

Thursday, October 27, 2022

Very Brief Blog: National Ads Support Delay of PAMA Lab Price Cuts

National ads support delay and revisions to PAMA, reducing lab price cuts.   The legislation currently in play is called SALSA.

Ads I see here in my home office on national media:

Wednesday, October 26, 2022

Brief Blog: Yet Another Article (This Time NEJM) on CMS, NCDs, CED (TCET)

 We're seeing a bonanza of articles on how CMS should make coverage decisions.

Ten days ago we had an article in JAMA by Fleisher and Blum of CMS on revising the NCD process, under the moniker TCET (Transitional Coverage for Emerging Technology) and perhaps doing more "CED" coverage with evidence development.  I link to it, and link to commentaries on it, here.

Also ten days ago, article in Health Affairs (linked to a longer, underlying white paper) by Tunis et al. on TCET, CED, etc.  Here.


When it rains it pours.  Today, an article in NEJM by Dhruva et al. on CMS and NCDs and CED - under the umbrella of the early 2022 "Aducanumab" decisions, but primarily to discuss CMS frameworks and future directions.  Worth reading.  Find it here.

Brief Blog: Personalized Medicine Coalition Summit, November 15, Boston; 48-page Booklet

Personalized Medicine Coalition has a webpage for its November 15, 2022, summit in Boston (with evening reception November 14).  It is titled, Personalized Medicine and the Patient, and they have released a 48-page companion booklet and program guide.

Find the home page here:

(See also an "Individualizing Medicine" summit in Rochester MD, November 2 - here.)

Find the 48 page PDF program guide here.

According to the registration page, the Summit live streams for free if I read it correctly (with registration).  In person rates are on a sliding scale from patient/advocate ($250) to member ($750) to non-member ($1000).

Brief Blog: FORCE Supports New Book on Hereditary Cancer

FORCE, the organization "Facing Our Risk of Cancer Empowered," supported publication of a new book from Johns Hopkins University Press:  "Living with Hereditary Cancer Risk," Steligo and Friedman, editors.

The book was released in September 2022 and is available in paper, hardcover, and ebook from Amazon.

FORCE initially focused on individuals positive for BRCA mutations, but expanded its focus to multiple hereditary cancers and genes in 2020.

Find the FORCE webpage for the book's release here:

Find at Amazon here.

Clipping from FORCE web page below:

An explosion of discoveries since the 2012 publication of our book Confronting Hereditary Breast and Ovarian Cancer has improved what experts know about and how they approach hereditary cancers. Some of the most notable advances include:

  • new technology that lowered the cost of genetic testing and improved the ability to search for mutations in many genes in a single test.
  • the discovery of mutations in ATM, CHEK2, PALB2 and other genes that have been linked to hereditary breast, ovarian, pancreatic and/or prostate cancer risk. Although less research has been conducted on these genes, more information has become known as more people have had expanded panel testing. Most of these genes now have expert guidelines for risk management.
  • increased understanding of the genes linked to Lynch syndrome.
  • research on more effective, less invasive options for managing risk.
  • the development of targeted therapies to treat cancers in people with inherited mutations found with genetic testing or tumor mutations found with biomarker testing.

In recognition of these changes, FORCE expanded our mission in October 2020 from focusing solely on people with inherited BRCA mutations to also serve people with Lynch syndrome and those with a mutation in lesser-known genes including ATM, PALB2, BRIP1, CHEK2 and others that also raise the risk for one or more cancers.

With so many important changes in the world of hereditary cancer, we needed a new guide that reflects our commitment to providing accurate, up-to-date information and resources to the expanded high-risk community. Living with Hereditary Cancer Risk: What You and Your Family Need to Know, does just that.

Sunday, October 23, 2022

Side Note: New 2022 Review, History of Transgender Surgery; Medicare Notes

The new October issue of Annals of Internal Medicine has a long historical article, and an accompanying Op Ed, on the history of transgender surgery in the US.   See Magrath WJ, 2022, "The fall of the nation's first gender-affirming surgery clinic," here.   See also Keuroghlian AS & Radix AE, "A cautionary tale: The doomed gender identity clinic at Johns Hopkins Hospital," here.   They discuss the history of the transgender surgery center at Johns Hopkins, which closed in 1979.


Not mentioned are the Medicare policy tie-ins, which I'll just annotate briefly.   


Medicare had no national policy on gender surgery until 1989, when it introduced a negative NCD, based on a extremely faulty "technology assessment" of the time (1980/1981).  (So this was a couple years after the Hopkins event and in Year 1 of the Reagan administration).  The negative NCD was published ;ater in 1989 as a few paragraphs in the Federal Register (August 1, 1989; Fed Reg 54:34555, at 34572.)  Here.  JPEG at bottom of this blog.


This 1989 NCD was thrown out by a panel of administrative law judges in 2013/2014, which I covered in a detailed blog in May 2014 here.  

(My 2014 blog notes that I had seen an optical character reading (OCR) version of the 1980/81 tech assessment, but the blog was edited by me to note that this had become a dead link in 2016. I don't seem to have a hard drive copy of that tech assessment, but in appears in bibliographies of academic articles on this Medicare history and in a discussion here.)   


Since the NCD was active 1989-2014, when it was thrown out, CMS had no national statement on transgender surgery for a couple years.  

In 2016, CMS released a draft, and then final, NCD, that said that transgender surgery was a local MAC decision.  (Blog from June 2016 here).  NCD 140.9 homepage here.  

While the draft and final decisions in 2016 both left coverage to the MAC level, there were  substantial redlines between the draft and final.  Click to enlarge:

2016 (redline draft v final)


Thursday, October 20, 2022

Peer Reviewed Study of Payor Coverage vs Guidelines (Wong et al 2022)

Earlier this month, I noted that California had vetoed a biomarker-access bill regulating insurance - here.  Bills of this type have been promoted by oncology community stakeholders like American Cancer Society Cancer Action Network (ACS CAN)

See a subscription deep dive article on the California veto at Genomeweb, here.

What I had not seen before, a May 2022 article by Wong et al., supported by ACS/CAN.  This article looks at payor coverage by state in comparison to guideline recommendations, and found that payor coverage often fell short of guidelines.

Find Wong et al. 2022 here:

Aim: Commercial plan coverage policies for multigene panel tests may vary and could result in geographic variation in coverage due to the fragmented nature of the commercial insurance market. This study aimed to characterize the alignment of multigene panel tests coverage policies to that of clinical guidelines, overall and by state. 

Materials & methods: We reviewed NCCN Guidelines® for four tumors. Public coverage policies were identified via web search. Payer policies included those with the largest or second largest number of commercial lives in each state. Policies were classified as ‘more restrictive’ or ‘consistent’ with the guidelines. 

Results: Of 38 plans/policies reviewed, 71% were classified as ‘more restrictive’ than the guidelines, with variation in the number of commercial lives by state. Among these, 52% restricted on panel size and 63% restricted in all or select tumors. 

Conclusion: Most coverage policies were more restrictive. Clinical guideline clarity and state policies may improve alignment to guidelines and geographic variations.

Saturday, October 15, 2022

MolDx: More on the Draft Melanoma LCD; Timelines; A Model LCD Request Letter Format

One of the issues with the multi-MAC MolDx system is that all four of its participating MACs must post draft LCDs, before they can be reviewed and finalized.  

In June 2022, I wrote about a complex MolDx LCD for molecular melanoma diagnostics (here, DL39345, released June 23).  It looks like this draft LCD was not posted until October 6 in the CGS MAC jurisdiction (here, as DL39389).  Comment til November 19, while there public meetings separately in KY and OH on October 25 and 26.

My earlier June blog provided a full discussion (DL39345).  Since then, AMP has submitted public comments opposing the requirement the test be ordered by a dermatopathologist only (here).  

Here, some additional observations.


I don't believe I noticed earlier, this draft LCD has a Request Letter provided.  

While the 14 page PDF has no date (or date redacted), I noticed the final title of the PDF was "Final 15 Jan 21" suggesting a review time of about 18 months between the request submission and the draft LCD's first posting in June (and 21 months, til the posting in the CGS MAC.)  

Generally, MolDx MACs post final LCDs 11, 12, or 13 months after the draft LCD.  

This suggests about 30 months (2.5 years, or longer, depending on the CGS timeline) between the submission letter and the final LCD (and, add 45 days for the final LCD to become effective.)

According to the 14 page request, Castle met with MolDx in October 2020 to help plan its January 2021 LCD submission and rationale.

Although there's no name or date on the letter, the splash screen for the issue description and requestor information says it's from Matthew Goldberg MD.  

"Read How a MOLDX LCD Request Letter is Written"

Find the 14 page request here:

Note the request letter is posted at CMS, not at a MAC.   A duplicate cloud copy here.

Foundational LCD and the Paradox of "Comparable to Covered Tests"

The LCD is a "foundational" or umbrella LCD that will cover future tests on a rolling basis if MolDx determines that they have "equivalent or superior performance to covered tests."

  • "Tests that demonstrate similar indicated uses and equivalent or superior performance to covered tests may similarly be covered under this policy."

However, this is trickier than it sounds.  

Many tests might be listed in the associated billing article only under the generic code 81479 (other molecular test), therefore, it would be hard for the reader to know what covered tests to look up, to compare their performance. 

This particular LCD article DA59163 covers unknown tests under 81479, plus 2 named tests 0090U, 0314U.  

These 0090U Myriad Mypath (now acquired by Castle) and  Castle DiffDx.  

0090U MyPath is an ADLT test at $1950.    

0314U is being priced.  A majority of CMS advisory panelists last summer recommended to crosswalk 0314U to 81529, which DecisionDx Melanoma, an ADLT test at $7193, one of the highest price tests on the whole lab fee schedule.   4 panelists recommended a crosswalk to 0090U, which CMS preferred.  This MyPath test is $1950, the crosswalk price being offered by CMS and far lower than $7193.

Nerd Note 00

The 14-page request letter attached to this LCD is a useful model.  A prior LCD had a request letter attached, but it was redacted to little more than a table of contents (here). 

Nerd Note 01

ADLT rules require that the test be made by the original developer, not simply licensed in, and not sold (unless a lab is acquired by a successor owner).  However MyPath 0090U appears to have been "acquired" by Castle.  And it's still listed as an ADLT.   This suggests that CMS applies the "not bought or sold" rule at the original time of deeming a test to be an ADLT, and CMS does not apply the rule again if circumstances change.  

(While in some places the rules are vague about whether a "lab" or possibly "a test" is being sold, others define this as "a successor owner of the laboratory" and not "of the test.")

Nerd Note 02

Why is 0314U Diff Dx being priced by CMS, why isn't Castle requesting it be an ADLT?  

According to the MolDx DEX Test Registry as of 10/2022, DiffDx is "not covered" so it can't apply to be an ADLT yet.  

Screen  shot: 14 page PDF LCD request letter

Friday, October 14, 2022

Very Brief Blog: in JAMA, Gottlieb & McClellan Promote VALID (FDA Diagnostics Regulation)

As numerous trade journals have reported over several months, the VALID ACT (FDA control of LDT diagnostic tests) has been rising and falling regarding its odds of becoming legislation.

In a new twist, Scott Gottlieb and Mark McClellan, both former heads of the FDA, strongly encourage Congress to pass the VALID ACT and as soon as possible.   (McClellan is a former head of both FDA and CMS).

Find it here:

Among others, AdvaMed is also urging Congress to push ahead with VALID (here).

The Regulatory Affairs Professionals Society, RAPS, hosts a two day course on the new European IVD regulations, October 18-19 in Amsterdam, full agenda here.

New Article: Tunis et al, Improving Medicare's Coverage of Innovative Technologies

Just yesterday, I wrote about a new article in JAMA by CMS leadership on improving Medicare coverage of new technologies.   Here.

Today, new news.  An article by Tunis and colleagues in Health Affairs, "Improving Medicare Coverage of Innovative Technologies."  Find the article here:

Note that the Health Affairs blog article links through to a longer, 33-page white paper online at Tufts, here.

The article opens,

After advances and setbacks, policy makers have revived efforts to improve the Medicare coverage process for new medical devices. Industry groups and others have long argued that the existing process is inefficient and unpredictable, and that a streamlined pathway for coverage decision making is needed for novel technologies that address serious and life-threatening illness. However, more rapid coverage processes, with shorter review periods and greater reliance on intermediate or surrogate endpoints, often mean more uncertainty about a technology’s risks and benefits at the time of coverage. 

The enduring challenge is how to expedite the process to ensure that Medicare beneficiaries have appropriate, timely access to novel medical technologies, while providing robust and efficient mechanisms for evidence generation in the pre- and post-coverage period.

Tunis et al. cite to Holtzman 2018, a review of parallel review.  They cite four products in the P.R. pathway; I was only aware of Cologuard 2014 and Foundation Medicine 2017.   Both Cologuard and Foundation Medicine reflected complex or specialized policy circumstances.  (That is, one-offs from which generalizations would be dicey.)   The Cologuard test, as a preventive benefit, was flatly non covered w/o a CMS NCD, and for CMS, it was a chance to show it could use P.R. at least once, the first use.  And Cologuard clearly performed a good bit better than FIT, so the chance that FDA would not approve it or CMS would not cover it was, I would think, infinitesimal.  (As an original scientific achievement, it's a big deal; as a regulatory achievement for CMS, it was a six-inch putt, once the data was in.)  And the Foundation Medicine NCD was originally, in its first draft, as much about cancelling all coverage of LDTs, as of allowing coverage of the Foundation FDA test.

However, P.R. was also used for a Medtronic renal denervation system that got stuck without FDA approval, and the Boston Scientific cardioverter w/o transvenous pacing leads (EMBLEM).(*)   
Holtzman 2018 Table 2

In their white paper, Tunis et al. cite three possible reasons for little uptake of P.R.: (1) Company concern it will delay market entry or trigger a premature negative NCD which wouldn't occur otherwise; (2) CMS will request too many changes to the FDA plan; or (3) NCD will trigger "CED" which otherwise wouldn't come to attention.   I would add a different and 4th reason: I'm aware of several companies that tried to get into the P.R. process and were turned away at the door (declined even a pilot meeting; the initial P.R. request is routed through FDA staff and got canceled there with a quick "Dear John" email reply.)

As an aside, when CED is discussed (as in Tunis et al.) it's not usually mentioned that some examples of CED, I believe, simply match up to FDA-required post registrational clinical studies (thus, not requiring or reflecting any additional expertise at CMS).  


(*) If I'm reading the 2017 decision on "leadless pacemakers" correctly, it requires CED but it doesn't mention parallel review, so even if I'd read it, I wouldn't have added it to my parallel review list.  I didn't see this on Google either; only in Holztman Table 2 line 4, above.


Thursday, October 13, 2022

Brief Blog: AMA Releases Bonanza of 35 PLA Applications

The AMA CPT has released for comment some 35 code actions for the October or Fourth Quarter PLA cycle.  A few of the applications are withdrawals or revisions, but most are for new PLA codes.

Find the AMA PLA home page here:

Find the AMA PLA PDF of the new action items here:

Comments on the PLA applications are offered only for a few days; see the PLA PDF for "Zendesk" instructions but do this right now if you want to comment on a code.

The PLA committee will meet and vote around November 7.   The AMA CPT committee will convene (quite briefly) by teleconference on November 21 to endorse the codes, so they can be released January 1.

Codes - A Few Highlights

It looks like the first agenda item is a PLA code for the Adaptive Biotechnologies FDA-approved CLONOSEQ test.

Two codes that just became active in July (0324U, 0325U) for cell culture drug response in oncology, are poised for deletion.  CMS had proposed to crosswalk these to the (I believe) off market Predictive Therapeutics cell culture codes from around 2014 (81535, 81536x3, $1100) whereas a near-identical test 0248U is in current gapfill at about $2600.

There is a code triplet for Oncuria (bladder cancer detection) and one for OVAWatch (ovarian cancer detection).   

There is a 17-pathogen bladder pathogen test (Qlear Urine), and a reflex version including antibiotic resistance genes.  There is a different test for "20 or more" pathogens in urine.  There is a 55-pathogen respiratory test. A separate wound infection test has 34 pathogens and 21 antibiotic resistance genes.  There are two complex kidney function tests (NaviDKD and PromarkerD).

There is the ArteraAI 128-image prostate image test with AI.

There is a somewhat complex (15 gene) cfDNA test for colorectal cancer.

Genesys has a 143-gene carrier panel.

Brief Blog: JAMA Article by CMS on CED/TCET. CMS Opens Registration for MedCAC on CED.

In the past month, Medicare calendared a December 7, 2022, MEDCAC (advisory meeting) on its Coverage with Evidence Development Paradigm, and AHRQ issued a report on CED to date.  

Now, CMS has published a JAMA article on coverage for early technologies, AND, opened registration for the December 7 webinar event.

Find the JAMA article here:

See trade journal coverage of the JAMA article here:

Find the Fed Reg announcement of the December 7 event here:

Find the registration for December 7 here (scroll down for reg)

AHRQ's draft report, which will be the core discussion document on December 7, is here:

From a separate workstream, at Tufts University, see a new paper on "Improving Medicare Coverage" by Tunis et al.

Regarding the JAMA Fleisher article, see comments below; I think there may be more content or at least as much value and material in a slightly later article by Dhruva et al in NEJM - here.


According to the new JAMA article, CMS is working on a regulation that would allow it to review data before FDA approval and determine a coverage pathway (such as "local coverage vs national coverage" and "coverage with or without CED").   Honestly, this description makes it sound awfully close to "Parallel Review" ahead of FDA approval, something that has barely, ever, been used in its 10 years of existence.   But we'll see what develops.  

Here are the four bullets of the proposed plan, as per JAMA:

The CMS rule that is being developed will meet the following principles:

  1. Manufacturers may enter the process on a voluntary basis. This process will be limited to medical devices that fall within the Medicare statute and are relevant to the Medicare population.1
  2. The CMS may conduct an early evidence review (before the device secures FDA marketing authorization) and discuss with the manufacturer the best Medicare coverage pathway, depending on the strength of the evidence collected.
  3. At the manufacturer’s request, CMS may initiate the coverage review process before FDA market authorization, which could require developing an additional evidence development plan and confirming that there are appropriate safeguards and protections for Medicare beneficiaries.
  4. If CMS determines that further evidence development is the best coverage pathway, the agency would explore how to reduce the burden on manufacturers, clinicians, and patients while maintaining rigorous evidence requirements.
The boldface red text both captures the key points of the JAMA plan, and, indicate areas that overlap with existing Parallel Review either explicitly or by common sense (obviously, CMS already would want to reduce, not raise, "unnecessary burdens," even if that phrase isn't explicit in parallel review).  


This events are the rolling out of a general program for technology coverage, which goes under the umbrella term "Transitional Coverage for Emerging Technologies" TCET.   CMS began holding town halls soliciting ideas on improving coverage last spring.   AdvaMed has a detailed white paper on the topic.    

I made a mid-September five minute video guide to what's happening -

Monday, October 10, 2022

Nerd Note: Statute that CMS Can't Apply Changes Retroactively

Updates 2023/03

Wolverton at AKIN law firm discusses.

Section 102 of H.R.3391 - Medicare Regulatory and Contracting Reform Act of 2001

I am not sure there is any matching regulation so 1871(e)(1) may be viewed as self implementing and may have no explicit regulatory discussing in the CMS CFR.

An example of CMS going to some detail and claiming retroactivity due to statutory instruction, under proper use of 1871 is found at:  3 FR 29699-711, Long term care hospitals, 5/22/2008.

From time to time a company has grounds to protest that an adverse change from CMS is being applied retroactively.  But generally, CMS's hands are tied from retroactive changes.

Here's the regulation.

Rules and Policy Statements Aren't Retroactive:

(e)(1)(A) A substantive change in regulations, manual instructions, interpretative rules, statements of policy, or guidelines of general applicability under this title shall not be applied (by extrapolation or otherwise) retroactively to items and services furnished before the effective date of the change, unless the Secretary determines that—

(i) such retroactive application is necessary to comply with statutory requirements; or

(ii) failure to apply the change retroactively would be contrary to the public interest.


This law 1871 also contains a clause, 1871(a)(4), that a final regulation must be a "logical outgrowth" of a proposed regulation, or else it triggers a new public comment period.  For example, CMS could propose a timeline be 12 or 24 months, and finalize at 18 (a logical outgrowth).   But CMS can't propose "Policy A" and finalize as "Policy Z" when Z is wholly different policy that is not a natural outgrowth.  


For a Supreme Court case hinging on the details of CMS regulatory processses like the quote above, see Alina 2019:

Readers might enjoy reading SSA 1871 as a whole. 


From Fed reg 2008, re a retroactive rule claimed by CMS with detail:

Section 1871(e)(1)(A) of the Act provides that a substantive change 

in regulations, manual instructions, interpretative rules, statements 

of policy, or guidelines of general applicability under this title 

shall not be applied (by extrapolation or otherwise) retroactively to 

items and services furnished before the effective date of the change 

unless the Secretary determines that (i) such retroactive application 

is necessary to comply with statutory requirements; or (ii) failure to 

apply the change retroactively would be contrary to the public 

interest. As explained in the paragraph above, the MMSEA requires the 

Secretary to implement various policy changes either contemporaneously 

with the enactment of the MMSEA on December 29, 2007 or beginning with 

cost reporting periods beginning on or after December 29, 2007 as 

applicable. Therefore, under the authority of section 1871(e)(1)(A)(i) 


[[Page 29708]]

the Act, we are making the provisions of this interim final rule with 

comment period that implement sections 114(d) of MMSEA retroactive to 

December 29, 2007. The statute also requires that section 114(c)(1) and 

(2) be implemented beginning with cost reporting periods beginning on 

or after December 29, 2007. Therefore, under the authority of section 

1871(e)(1)(A)(i) of the Act, we are making the provisions of this 

interim final rule with comment period that implement section 114(c)(1) 

and (2) effective for cost reporting periods beginning on or after 

December 29, 2007. Additionally, as explained previously, the Secretary 

also finds that it would be contrary to the public interest if these 

provisions were not made effective on December 29, 2007 or for cost 

reporting periods beginning on or after December 29, 2007, as indicated 

above. Therefore, under the authority of section 1871(e)(1)(A)(ii) of 

the Act, we are making these changes effective under the timeframe 

noted above.

    For the same reasons noted above, we find good cause under section 

553(d)(3) of the APA to waive the 30-day delay in effective date.



Thursday, October 6, 2022

California Vetoes Bill that (Would Have) Required Biomarker Coverage

Several states, including Illinois and Louisiana, have passed bills that mandate guideline-recommended or FDA-approved biomarker coverage.   The bills vary; some are specific to cancer and some are not.

One of the most recent proposals has been vetoed by the Governor of California, after having been passed by the legislature at the end of August.

The open access news story is available at Genomeweb.

The governor's online rationale is here.

Press release from American Cancer Society here.  Univ Calif had supported the bill.

See a related study from ACS in May 2022 regarding patients' access to biomarkers and insurance coverage, here.

See the ACS Cancer Action Network home page of press releases related to access to cancer care here.

See also: and

Tuesday, October 4, 2022

Very Brief Blog: Two Interesting Articles in JACR

 I ran across Journal of the American College of Radiology around 2010, working on CMS PET scan policy, and I've been a fan ever since.   While lab and pathology journals (e.g. AMP's Journal of Molecular Diagnostics) have important policy articles from time to time, JACR is mostly policy and business articles, some of them really intriguing.  I've always assumed that there's crossover between imaging and lab diagnostics, both being, diagnostics.  I mused on this in March 2022 here.

Here are two interesting articles from the September issue, and if I read correctly, both are flagged as open access.

In a really unique eight-page article, Stefan Tigges MD presents diagnostic test metrics (NPV, etc) in a sophisticated way yet using a cartoon format.  Find it here.

On a wholly different topic, Brandser & Kothari talk about the pressure of unpredictable work volume and reading queues in day to day radiology, and how their group practice came up with some novel work plans to improve effectiveness.  (Some of the fastest readers in the group, were happy to get paid extra to do "bunker shifts" reading a fixed aliquot of images in queue.  And during that time - say, on a Saturday - having no other duties, and just working from home.  Sounds simple, but someone had to think of it.)  Sort of like an attorney going into the home office on Sunday and knocking out 1.5 billable hours. 

One thing I noted - while CMS doesn't have a fixed time for physician RVUs, they are often around 3 per hour.   E.g. a 45-60 minute office visit, gets 2.6 work RVUs.   This radiology practice has a standard metric of an output of 10 work-RVUs per hour.   (For an internist, that would be 4, "45 minute" office visits per hour to log 10 w-RVU per hour).  (It's like the joke about a procedure that gets on the books 37 or 42 CMS minutes, but is scheduled every 20 minutes at all clinics in the real world).   

It's a Tough World for Tests: NGS MAC LCD on Ovarian Biomarkers (Nothing Covered)

There are a number of biomarkers for detecting ovarian cancer or working up adnexal masses.

If you'd like to be prepared for a tough review picking apart your new diagnostic test, get hardened up by reading the NGS MAC article about ovarian cancer tests (A58112) and the LCD (L38371).  I just ran across these documents; they're from 2020.

My only point here, is when you enthusiastically bring your shiny new test and your one first publication in 40 patients to Medicare or another payer, you may get some pretty rough handling.  (Many startup labs don't foresee this.)

Summarizing the NGS MAC author:

  • "This is a non-coverage policy for all multi marker serum tests related to ovarian cancer testing."

Read the LCD and data review here:

Read the public comment article and the NGS rebuttals, here:


While not the only tests reviewed, two of them, Ova1 and OVERA come from Aspira Women's Health.  Both have been FDA cleared.

Nasdaq AWH


The same MAC has a more liberal policy for several biomarker tests, like 4KScore, which are used to guide management of an intermediate PSA like "4".  It was just updated to add EPI, MyProstateScore, and isoPSA.

The summary coverage for prostate is below.  "Coverage" decisions are obviously longer than non-coverage decisions; the latter are basically one word, "no."

Coverage Indications, Limitations, and/or Medical Necessity

One biomarker test, ordered by a physician or other qualified health care professional (i.e., NP, CNS, PA) is covered ONCE per year. For men >/ 45 years old receiving testing prior to potential biopsy, covered testing includes %fPSA, PHI, Select MDx, 4K Score or MyProstate score, in men >/ 50 covered testing additionally includes EPI and isoPSA in those men >/ 50 with a PSA > than 4ng/ml. Those men, who need a repeat biopsy in the setting of patients thought to be at higher risk despite a prior negative biopsy, covered testing includes %fPSA, PHI, 4K Score, PCA 3, Confirm Dx, MyProstate Score and isoPSA with confirmed * moderately elevated PSA (>3ng/ml and <10ng/ml; or PSA >/4ng/ml and < 10ng/ml in men > 75 years of age) with BOTH of the following... [high risk indication of cancer, or contraindication to biopsy anyway]