The Fast Summary
Novitas released a long-awaited massive omnibus LCD for cancer testing. It includes (1) very complex rules, (2) a list of non covered tests (including from some publicly-held labs), and (3) a very very long "billing article."
++++++
What Happened?
Rarely, an LCD or NCD is so talked-but that it's known by its number - "NCD 90.2," the CMS NGS NCD, being one example. A nominee from the LCD side is Novitas' massive and important LCD for oncology - L39365.
I should say, it's massive "now" - the draft was 14 pages, 6000 words, and the final is 68 pages, 30,000 words.
There are now 230 citations (up from only 19).
A lengthy section of definitions, some quirky, have been added. A billing article runs hundreds of pages. In all about 400 pages of reimbursement documentation was just released!
- Billing & Coding, A59125, 283 pages.
- LCD, L39365, 68 pages.
- Response to Comments, A59417, 44 pages.
- I get 395 pp.
June 2022 to June 2023
Last June, 2022, Novitas put out a major revision of this complicated LCD. I would say, charitably, that it "defies easy summary" even in its original 14 page form. But, it was founded on a heavy reliance on NCCN guidelines and similar sources for its coverage rules.
Novitas went right to the wire on this one, chased by the legal one-year deadline that CMS imposes between a draft and a final LCD.
A Rare One-Off Release Date
For one oddity, whereas LCDs normally are issued nationally on Thursdays - e.g. June 1 - this one came out on Friday, June 2, after market close. Genomeweb here.
Look It Up!
Find the 2022 draft version here and its tracking sheet here.
- The revisions include, "literature reviews added for 13 specific tests (Cxbladder, ThyroSeq® CRC, PancraGEN®, DecisionDX-Melanoma, DecisionDX-SCC, UroVysion® FISH, Colvera™, PancreaSeq® Genomic Classifier)."
- And also, "Updates to the Analysis of Evidence including Technology Assessment of knowledge bases and literature reviews added for 13 specific tests (Cxbladder, ThyroSeq CRC, PancraGEN, DecisionDX-Melanoma, DecisionDX-SCC, UroVysion FISH, Colvera, PancreaSeq Genomic Classifier)."
The responses to comments are here, A59417:
https://www.cms.gov/medicare-coverage-database/view/article.aspx?articleId=59417&ver=17
The FINAL LCD, L39365 2023, is HERE:
https://www.cms.gov/medicare-coverage-database/view/lcd.aspx?lcdId=39365&ver=80
The billing and coding article is here, A59125:
https://www.cms.gov/medicare-coverage-database/view/article.aspx?articleId=59125&ver=26
Tier 2 Codes Handled?
Novitas has paid out a billion dollars in the last couple years for Tier 2 codes, at least a good part of which, to labs indicted or pled guilty in court.
In this billing article, only two of the Tier 2 81408 genes are eligible for coverage, which are ataxia telangiectasia and neurofibromatosis. Folks requiring either of this two genes will be rare as hen's teeth in the nursing homes of Texas and Florida. Novitas has paid zillions of dollars for 81408 in multiples of 2 per patient per day, for several years (!!!!!) It's impossible to imagine anyone needed both ataxia telangiectasia and neurofibromatosis testing on the same day for $4000 again.
81493 - Proteomic Gone
I don't see (by search) a reference to 81493, the CardioDx test I recently wrote about. It doesn't seem to be in Group 1 of this LCD, where it had been just recently. I'll keep looking. It might be dropped for being a "proteomic" test as the LCD re-emphasizes it is for DNA/RNA tests aka "genetic" tests (for Novitas).
81455 - CGP
The LCD doesn't really discuss one of the biggest areas in oncology, large tumor gene profiles. In one place, the billing article says these are covered "when they meet the criteria of the LCD," as if the provider is suppose to guess at that, by reading the 68 page LCD. However, sections 7 and 8 seem to provide nearly unlimited coverage of CGP as long as the patient has a cancer diagnosis code in the solid tumor series (#7) or the hematopoeitic series (#8).
Hot topics like liquid biopsy in lieu of paraffin, or MRD testing for relapse, or MRD testing for drug response, don't seem to be mentioned. In contrast, in the MolDx policy system, these are among the most actively and vigorously discussed and updated areas.
Large Changes in Content, w/o Public Review
The increase in LCD content, from about 14 pages to 68 pages or 5X, means that much of the text that is now effective on July 13, 2023, was not prior reviewed by the public or commented on by the public.
When CMS issues federal regulations, it must renew a comment period on any final regulation that is not "a logical outgrowth" of the proposal. SSA 1871 (a)(4). However, that is a law about "regulations" not "LCDs."
My own draft-to-final comparison or redline is in the cloud here.
Not Medically Necessary
This section of the LCD is clipped and pasted below.
Limitations.
The following are considered not medically reasonable and necessary:
- DecisionDx-Melanoma*
- DecisionDx-SCC* [see 5/23/2022 press release re Pittsburgh investments]
- Cxbladder Detect*
- Enhanced Cxbladder Detect*
- Cxbladder Monitor*
- Cxbladder Triage*
- Enhanced Cxbladder Triage*
- Cxbladder Resolve*
- [BQ: Discussion of both Castle and CX products are about 5000 words long. Each has one initial discussion called "evidence" and a later discussion called "evaluation." Clipped here, here. For example, the statement "The study has 100 patients and 80% accuracy" is "evidence" while the statement "80% accuracy is too low for coverage" is "evaluation."]
- Colvera*
- PancreaSeq Genomic Classifier*
- PancraGEN*
- ThyroSeq CRC
- UroVysion fluorescence in situ hybridization (FISH)*
- A genetic test with unestablished analytical validity, clinical validity, and/or clinical utility.
- Interventions with levels of evidence not identified by ClinGen24, NCCN25, or OncoKB26 as demonstrating actionability in clinical decision making OR interventions that are non-covered per MAC review.
- Genetic testing in patients who do not have either an established diagnosis of cancer or substantiated suspicion of cancer as determined by a clinical evaluation and abnormal results (cancer or suspicious for cancer) from histologic, cytologic, and/or flow cytometric examination. (Except where otherwise specified in the Covered Indications as reasonable and necessary) (See SSA Section 1862(a)(1)(A))
- Genetic testing of asymptomatic patients for the purposes of screening the patient or their relatives. (See SSA Section 1862(a)(1)(A))
- Repetitions of the same genetic test on the same genetic material. (see Medicare NCCI Policy Manual, Chapter 10, Section A Introduction)
See also additional comments and rules in the lengthy billing article.
The LCD - A "Quant" View
It took me a while to wrap my head around what's going on in this lengthy document. I cut/pasted parts into Word for this analysis:
|
01 |
Definitions |
4 pp, 1680 words |
|
02 |
Coverage Rules & Authorities |
9 pp, 1800 words |
|
03 |
Authority Sources Analysis |
17 pp, 7400 words |
|
04 |
Specific Tests: Evidence & Analysis |
40 pp, 19,000 words |
|
|
|
Of the above, first 8 pp and 4500 words are “Evidence,”
the rest, “Analysis” |
|
05 |
Bibliography – Final |
17 pp, 7250 words, 230 citations |
|
06 |
Bibliography – Proposed |
3 pp, 520 words, 19 citations |
Wild Swings
Numerous payers (Medicare or commercial) issue policies that one or another MAAA test isn't yet proven essential. So that's not surprising in itself. However, it seems to me the author of the LCD takes some pretty wild swings that are too picky or simply odd to make sense scientifically in the context of an HTA. Often generalistic paragraphs about measuring the cell biology of RNA are cut-and-pasted repeatedly into different product reviews in this LCD.
Can AI Help Us Understand This Massive LCD?
Can AI help understand this? Maybe. In a separate appended blog, for the AI-curious, I ask GPT4 a half dozen questions about different parts of the L39365 LCD.
https://brucedocumentblog.blogspot.com/2023/06/using-ai-to-help-understand-novitas-lcd.html
