Header: AMA continues to work on revisions to "Appendix S" of the AMA CPT, which covers predominantly digital services.
The AMA held a discussion in September during the CPT meeting, then a fly-in (+virtual) meeting December 8. It released versions A and B in December. Now on February 3 (just prior to a February 5 meeting) AMA releases quite a few additional revisions.
At which point I slow down, as the exact content of the revisions are under confidentiality.
Here are some cloudy generalities, as seen by an AI service, Chat GPT 5.2 [start]:
____________________
At a very high level, Appendix S is CPT’s attempt to create a shared vocabulary for describing what “digital services” do in relation to clinical decision-making — whether software is merely highlighting information, deriving new parameters, or moving into interpretive and action-oriented roles.
In December, the early “A” revision largely focused on sharpening definitions and boundaries: clarifying what counts as meaningful analytic output, how categories differ conceptually, and where simple data handling ends and clinically relevant digital processing begins.
The subsequent “B” revision took a more structural turn, paying closer attention to how these digital functions might relate to existing CPT code architecture, including issues around whether a digital function is integral to another service or potentially distinct, and acknowledging that software increasingly changes how clinicians interact with information rather than simply adding traditional interpretive work. "Examples" were dropped, however.
What's New - with 48 Hours to the Meeting
The February “C” revision moves further in that direction, giving more explicit attention to the relationship between digital services and physician work — including performance, interpretation, and responsibility — while also trying to stabilize the framework so it can be applied consistently across different technologies and risk levels. In effect, it pushes Appendix S from a conceptual taxonomy toward a more operational set of principles about how digital outputs relate to clinical use and accountability.
Never Fire-Tested Against Genomics
What remains notable, however, is that these principles have not been stress-tested in domains like genomic testing, where CMS does not treat the activity as physician work at all.
Genomics operates under laboratory fee schedule conventions, where software (including AI) is often part of the assay itself rather than an overlay on a RVU-based, billable physician service. As a result, Appendix S is evolving within a physician-service mental model that fits some fields naturally, but no one has started to reconcile it with areas like genomics that sit completely outside the AMA's physician services framework.
Parts of Digital Pathology Migrate from Physician Service to Machine-Learning Event
Importantly, future forms of digital pathology may begin to resemble this AI-genomics model more than traditional anatomic pathology. Image data will be transformed directly into machine-generated clinical outputs — for example, recurrence risk scores on a 1–100 scale — rather than supporting a pathologist’s experience (visual impressions and narrative interpretation.)
As that shift occurs, the gap between Appendix S’s physician-work assumptions and the operational realities of data-driven laboratory diagnostics may become even more pronounced.
Early Reports: MCED Medicare Legislation Passes Both House & Senate in Budget Bill.
The recent draft federal budget legislation included a section for Multi Cancer Early Detection MCED coverage, "IF" CMS were to write a national coverage determination (NCD) that allows it. The MCED section, 6221, is reported to have "survived" into the final edit, and has passed House and Senate by Feb. 2, 2026. See a rapid press release from Exact Sciences here.
The legislation allows CMS to write an NCD if it chooses, for an FDA approved MCED test. CMS is not required to wait for USPSTF endorsement. The benefit is guarded; for Medicare patients age 50-64, and price capped at about $600 (= Cologuard), and not starting for several years (and, if, CMS makes that NCD). Legislation for MCED has circulated in varying forms for a couple years.
##
See press from Jan 22, House passage. Here. Rep. Terri Swell (AL); the "Nancy Gardner Sewell Medicare MCED Screening Coverage Act." (Prior introduction as HR 842; added to Appropriation bill.)
See early support from American Cancer Society Cancer Action Network ACS CAN here.
While a Cepheid GeneExpert test was a top Part B code in 2024 (here), most new technologies for infectious disease, including point-of-care, have had a slow slog. One barrier is that CMS bundles payment for these tests in the ER setting (where CMS only pays for human DNA-RNA tests).
According to a report in 360Dx, T2 Biosystems will wind down operations. They note that trading of stock suspended as far back as February 2025.
360Dx writes, "The firm said it has "extremely limited cash and significant liabilities" and has concluded that "a wind down is the only viable path forward." ...Its core technology for pathogen identification and antimicrobial resistance testing directly from blood samples relied on magnetic resonance imaging and PCR.
Along the way, in summer 2019, its T2 Bacteria Panel won NTAP (inpatient add on payment) for FY2020. The add-on to the DRG (which is often $5000-15000) was $97, which applied only to the hospital's Part B FFS inpatients.
Header - Natera files SIGNATERA test with FDA. Keynote application is muscle invasive bladder cancer.
###
News item assembled by Chat GPT from newswires. Natera press release here. Oncodaily here. Urology Times here.
###
Natera has submitted a Premarket Approval (PMA) application to the U.S. Food and Drug Administration for its personalized blood test, Signatera CDx, seeking clearance as a companion diagnostic to help guide treatment decisions in patients with muscle-invasive bladder cancer (MIBC) following surgery.
The company is asking FDA to approve the test for detection of molecular residual disease (MRD) — traces of cancer detectable only at the molecular level — in order to identify which patients are most likely to benefit from adjuvant treatment with the immunotherapy drug Tecentriq (atezolizumab).
The filing is backed by results from the large, randomized phase 3 IMvigor011 trial, which tested whether circulating tumor DNA (ctDNA) detected by Signatera could be used to direct therapy.
In that study, patients whose blood tests showed ctDNA — indicating residual cancer after cystectomy — experienced significantly longer disease-free and overall survival when treated with atezolizumab compared with placebo.
Conversely, patients who remained ctDNA-negative had a low rate of recurrence without immunotherapy, suggesting many could safely avoid treatment and its potential immune-related side effects.
Together, the data support the idea that MRD detection can function as a predictive biomarker, not merely a risk signal, by identifying who is likely to benefit from therapy.
If approved, the test would move MRD detection in bladder cancer from an evidence-supported clinical strategy into an on-label clinical decision tool formally tied to a drug. That would mark a shift toward “recurrence-stage precision oncology,” where treatment decisions are based on molecular evidence of residual disease rather than tumor stage alone. Natera has characterized the application as potentially leading to one of the first FDA-approved MRD blood tests used to guide treatment decisions in a solid tumor setting.
MRD diagnostic tests have a longer history with blood cancers.
Taken together, the PMA submission signals that MRD-guided therapy in MIBC is moving from late-stage clinical research toward potential regulatory reality. The central question now is not whether ctDNA can stratify risk, but whether regulators agree it should be used formally to direct treatment decisions — a decision that could shape the future of MRD testing across many solid tumors.
###
Sidebar: Timing
The PMA pathway is FDA’s most stringent device review. Although the agency’s formal goal is 180 days after filing, actual timelines vary: straightforward applications may finish in under a year, while more complex ones — particularly In this case, review timing may also be influenced by the drug side of the equation. Because Signatera is being evaluated athose involving novel biomarkers, predictive claims, or companion diagnostic labeling — often take 10–14 months or longer because the review clock stops during sponsor responses. An advisory panel, if convened, can add further delay.
s a companion diagnostic, its labeling and approval are functionally linked to the clinical context and regulatory status of atezolizumab in this specific bladder cancer setting. Coordination between device and drug review teams is common in such cases, and alignment on indications, patient selection language, and clinical evidence can affect the pace of the process.
###
Sidebar: Tecentriq.
Tecentriq (atezolizumab) is a monoclonal antibody that binds PD-L1, blocking its interaction with PD-1 and B7.1 receptors on T cells. This releases an immune “brake,” restoring T-cell–mediated anti-tumor activity so the immune system can attack cancer cells.
I asked Chat GPT 5.2 to review and summarize a 50-minute dialog between RFK Jr and podcaster Katie Miller. Find the whole podcast here.
[Chat GPT 5.2]
Overview
In a wide-ranging conversation with podcaster Katie Miller, HHS Secretary Robert F. Kennedy Jr. frames his “MAHA” agenda around two headline moves: new dietary guidance urging “real food” while targeting ultraprocessed products, added sugars, and sugary drinks; and a revamped childhood vaccine schedule emphasizing optionality and clearer access to underlying evidence. He argues that industry influence has distorted nutrition policy and that vaccine safety surveillance and communication need rebuilding. Kennedy also touts healthcare price transparency and an AI-enabled overhaul of FOIA access. Personal notes—sauerkraut, a carnivore-leaning diet, 12-step meetings—underscore his health-focused persona and offers his rationale and priorities for 2026.
Why the Silence on Medical Innovation & Medicare Policy Matters
Notably, across his wide-ranging discussion about the future of U.S. health, the Secretary makes no reference to Medicare reimbursement, molecular diagnostics, genomics, precision medicine, or medical technology innovation. For stakeholders in diagnostics, biotech, and laboratory medicine, this omission is informative. It suggests that the current health-policy narrative being advanced at HHS is centered on population-level drivers of disease (diet, environment, lifestyle), institutional transparency, and vaccine policy, rather than expansion of high-technology clinical care. That framing does not oppose innovation outright, but it positions advanced diagnostics and precision medicine as secondary toupstream behavioral and environmental reform, which could influence future emphasis in funding, regulatory attention, and policy bandwidth.
On February 1, 2026, the NYT medical puzzle was a woman with abdominal pain. I gave Chat GPT 5.2 the first half of the article, then asked for a diagnosis. It gave one (the correct one.) I then gave Chat GPT the rest of the NYT article and Chat GPT provided a closing summary.
I provide links to the whole dialog and I also provide a summary of the story, below.
The past month, I saw a nice white paper by LabCorp about "Bridging testing gaps by enabling an FDA cleared IVD tumor profiling precision oncology test within your local community setting." Find it here.
On a similar theme, in December 2025, CAP Today published a lead article on, "How two labs took on in-house sequencing." Find it here.
I asked Chat GPT 5.2 to contrast and compare the two pieces.
MolDX and Z codes have been around for well over 15 years now, preceding by several years the new AMA CPT genetic coding system of 2012-2013.
How many Z codes are there? We can start to ballpark the number by going to the Dex Diagnostics Exchange Register, registering as a public user, and looking at the two catalogs. These are the TEST catalog and the "LABS & MFGRS" catalog.
The PAMA legislation of 2014 requires OIG to issue an annual report on lab spending - in part to show Congress the impact that the new "triennial market price reset" was having on lab spending. As Medicare nerds know, the reset occurred only once (in 2018) based on data from 2016.
My blog September 2025.
CMS released most of this data in September 2025, and I reviewed and discussed it then. Blog.
Genomeweb Discusses OIG.
Genomeweb/360Dx (Adam Bonislawski) covers the OIG report, and also has a long discussion of the weirder aspects of molecular spending.
What OIG found.
The report has extensive tables and graphics focused on genomic vs conventional lab spending.
The lab report suggests that the increase in genomic tests was due to spending on cancer, infections, and epilepsy.
The growth in cancer spending was national, and included MolDx states with big-name cancer genomics labs.
The growth in infectious and epilepsy was more dubious, occuring only Novitas states and mostly due to an nonspecific PCR code for pathogens (87798), the highest cost line item at $442M.
Epilepsy 81419 - a panel used only in non-NGS MAC and non-MolDx states - 81419 came in at $73M.
So for a few years Novitas had about the highest-paid code 81408 rare gene (sic) full length sequencing, circa $500M, and that fell to zeroat the same time as87798, uncontrolled pathogen code, role to circa $500M. Sigh.
Payments outside Novitas hot spots like Texas and Florida was very small, and billing for 87798 and 81419 by known labs (whether Quest-LabCorp type or GeneDx-Ambry type) was about nil.
The highest paid genomic code, 87798, nonspecific pathogen, was $442M, much of it "odd."
The second highest-paid code was 81528, Cologuard, at $311M. The third-highest code was Candida 87481 $102M.
The fourth and fifth codes were held by Guardant (0242U, 0326U, both circa $100M) and Decipher Prostate (81542) came in just behind at $95M, about the same as 0340U (Signatera MRD) and 81519 (Oncotype Dx).
On January 26/27, 2026, the big news was that CMS has announced there is virtually no change in rates for Medicare Advantage in CY 2027. They issued a press release, FAQ sheet, and a 160-page policy document.
Read here for Reuters. Read here for HealthcareDive. (HCD also here for comments from administrator Chris Klomp.) Or see picture for the main takeaway:
But wait a second - back on November 28, 2026, they issued a 137-page Proposed Rule for Medicare Advantage in 2027.
How to do the policy making processes differ?!?
I asked Chat GPT.
### SUMMARY OF THE TWO DOCUMENTS
CMS released two different CY 2027 “Part C & D” documents, but they serve separate legal functions — one changes program rules, the other updates payment mechanics.
The November 2025 Federal Register Proposed Rule (CMS-4212-P) is formal regulation. It amends the Code of Federal Regulations and covers policy areas such as Star Ratings methodology, marketing and TPMO oversight, enrollment processes, SNP policies, and Part D benefit redesign. This rule determines what Medicare Advantage and Part D plans must do operationally. It will later become a Final Rule.
The January 26, 2026 Advance Notice is not a regulation. It is an annual payment methodology notice required by statute. It updates MA benchmarks, growth rates, risk adjustment models (CMS-HCC and RxHCC), normalization factors, coding intensity adjustments, and Part D benefit parameters. It determines how CMS calculates plan payments, and becomes the April Rate Announcement, not a Final Rule.
###
### LINKS
Fed Reg November 2025 here. Advanced notice 160pp January 2026 here. Fact sheet here. January 2026 press release here.
###
###
You are absolutely justified in being confused — CMS uses two totally different legal pipelines that both talk about “Part C & D changes for 2027,” but they do different jobs.
Here’s the clean mental model.
🧠You are looking at two separate rulemaking universes
November 28, 2025 document
January 26, 2026 document
Federal Register Proposed Rule (137 pages)
Advance Notice (160 pages)
Docket: CMS-4212-P
Not a FR rule — payment notice under SSA §1853
Amends regulations (42 CFR Parts 422 & 423)
Updates payment formulas, rates, and risk models
Becomes a Final Rule later
Becomes the Rate Announcement (April 2026)
Legal authority: APA rulemaking
Legal authority: MA rate-setting statute
Changes program rules
Changes how plans get paid
📘 1. The NOVEMBER rule = PROGRAM POLICY RULE
This is a classic CMS regulation. It changes the rules of the game.
It covers things like:
Part D benefit redesign codification (IRA changes)
Star Ratings methodology changes
Marketing / TPMO oversight
Enrollment rules
SNP policy
MLR / reporting ideas
Network adequacy process changes
It literally amends the CFR. This is about what plans must do.
Source: The November rule states it revises “Part C and Part D regulations” and includes policy areas like Star Ratings, marketing, enrollment, and drug coverage.
2025 CMS 1128 90FR54894 137p Pr…
👉 Outcome: Final Rule later in 2026.
💰 2. The JANUARY document = PAYMENT & RISK ADJUSTMENT NOTICE
This one is not a regulation. It does not amend the CFR.
It exists because the Medicare statute requires CMS to annually publish:
MA growth rates
Benchmarks
Risk adjustment model updates (CMS-HCC, RxHCC)
Coding intensity adjustment
Normalization factors
Part D benefit parameters
Payment methodology details
It is explicitly called:
“Advance Notice of Methodological Changes for CY 2027 MA Capitation Rates and Part C and Part D Payment Policies”
2026 CMS 0126 Part C D 2027 Pro…
This is about how CMS calculates plan payments, not beneficiary rules.
👉 Outcome: Becomes the Rate Announcement (April 6, 2026), not a Final Rule.
🔑 The Key Distinction
Topic
Lives in November Rule?
Lives in January Notice?
Marketing restrictions
✔
✖
TPMO oversight
✔
✖
SNP eligibility policy
✔
✖
Part D benefit structure (legal codification)
✔
✖
MA benchmark math
✖
✔
Risk adjustment model calibration
✖
✔
Coding pattern adjustment
✖
✔
Star Ratings payment impact
✖ (policy only)
✔
Part D risk model updates
✖
✔
🎯 Why CMS does this (historical reason)
Medicare Advantage has two separate legal mechanisms:
A. Program rules
Made through APA rulemaking → CFR changes (That’s your November document)
B. Payment formulas
Made through annual rate notice → Actuarial / statutory process (That’s your January document)
This split goes back to the Medicare Modernization Act (2003) and earlier managed care law. MA payment is treated like insurance rate-setting, not standard rulemaking.
⚠️ Why it feels overlapping
Both documents mention:
Star Ratings
Part D
Risk adjustment
“Part C and D policies”
But:
November = WHAT the program requires January = HOW CMS calculates money
Example:
November rule: May change Star Ratings measure set. January notice: Applies Star Ratings to determine bonus percentages in benchmarks.
They run in parallel every year and converge when plans bid.
If you’d like, next I can:
A) Map how a single issue (like Star Ratings) flows through BOTH pipelines B) Show which parts of the January notice plans actually care about most C) Explain how Part D IRA changes appear in each document
Your confusion = totally rational. CMS’s architecture is… vintage 1997 managed care law meets modern Medicare.
#######
Biden year 4 Med Adv policymaking had many pages about insuring coverage parity, e.g. validity of LCDs in Part C. Trump Admin went radio silent. No change, I assume?
####
Short answer: No — neither document meaningfully addresses traditional coverage policy (NCDs, benefit categories, Part A/B service definitions, or Part C “same benefit” scope issues).
You’re picking up on a real shift.
📘 November Proposed Rule (CMS-4212-P)
This rule is about MA/Part D program operations, not Medicare coverage law. Its focus areas are:
Part D benefit redesign mechanics
Star Ratings methodology
Marketing / TPMO oversight
Enrollment and SNP policy
Reporting, MLR, network adequacy processes
It amends 42 CFR 422 & 423, but does not reopen:
National Coverage Determinations (NCDs)
“Reasonable and necessary” standards
Part A vs Part B benefit definitions
MA “uniformity” or “same benefit as FFS” doctrine in any substantive way
Even the IRA Part D sections are benefit design financing, not coverage scope. Source: The rule’s purpose is described as revising MA and Part D regulations on Star Ratings, marketing, enrollment, drug benefit structure, etc. — not coverage policy.
💰 January Advance Notice
This one is even further removed. It is a payment mechanics document:
Benchmarks
Growth rates
Risk adjustment models
Coding intensity adjustment
Part D bid parameters
It explicitly describes itself as updating capitation rates and risk models, not benefits.
No discussion of:
Coverage standards
Clinical benefit categories
Concordance with FFS NCDs
Scope of Part A/B services
🎯 What’s changed vs prior CMS eras
You’re right — prior administrations (especially 2021–2024) used MA rulemaking to discuss:
Guardrails around MA “reinterpretation” of coverage
FFS-equivalence expectations
Clinical evidence standards
Prior auth alignment
That language is largely absent here. The focus is:
Program integrity, payment calibration, and administrative simplification not Clinical coverage policy
🧠Bottom line
Topic
November Rule
January Notice
NCDs
❌
❌
Part A/B benefit definitions
❌
❌
“Reasonable and necessary”
❌
❌
MA must mirror FFS benefits
Only background law, not policy
❌
Coverage expansions or restrictions
❌
❌
These documents operate in the MA/Part D administrative and financial domain, not the Medicare coverage doctrine domain.
Your observation = correct: coverage policy discussion has moved offstage in these CY 2027 materials.
Stephane Budel and DeciBio post a four-page summary of some of their recent research on the moving l ine between tools and clinical diagnostics. Find their post and PDF here.
Writing,
Life sciences research tools (LSRT) companies are pivoting hard toward clinical and translational value pools:
- Illumina: ~60% of consumables now clinical...
- 10x Genomics: launching a CLIA lab...
- [more examples]
Adding, “clinicalization” requires a significant business model transformation. Reimbursement, quality systems, go-to-market, partnerships… it all changes...
It seems obvious to me that "value based care" and "managed care" won't solve the US outlier position for healthcare costs. But we've got a headline article in Health Affairs that makes the point.
See January 23, 2026:
Value-Based Payment And Managed Care Will Not Solve The Affordability Crisis
Among the unsurprising data shown here, Bond and colleagues in 2025 JAMA found that ACOs saved $300/pt after six years. Ryan et al. point out that in the target year, 2019, that was $500M savings compared to $800B spending - a small fraction of 1%. Open access and good reading.
The authors give some airtime to concerns that technology drives healthcare costs. We could go vastly farther in encouraging technologies that save healthcare costs, instead of making them almost impossible to adopt.
(For one promising example see PaigePredict here. I'm not promoting this particular example, but rather, it stands-in for a whole class of beyond-the-horizon breakthrough efficient technologies that deserve attention, also here.)
___
I find this unsurprising. If you go to capitated managed care, there are immediately huge incentives to "under-provide." If you call it 'value based payment," you are either selectively a trivial number of healthcare parameters (maybe 15, like flu shots, and heart failure readmission rates, and A1C), or to really track health care "value," you'd have an inordinately complex parameter system that doesn't exist.
___
The article doesn't have an abstract but here's a 200 word summary [Chat GPT 5.2]:
Policymakers keep backing value-based payment (VBP) as the fix for U.S. health care costs, shifting from fee-for-service to risk contracts like ACOs and Medicare Advantage.
But decades of experience show minimal or no real savings. Even the most optimistic recent ACO studies suggest Medicare savings of only 0.06–0.13%, and those estimates are likely inflated by patient selection and diagnostic upcoding, not true efficiency. Meanwhile, CMS bonus payments and gaming may mean net losses.
Claims that VBP indirectly slowed national spending growth (“spillovers”) don’t hold up. The U.S. slowdown began before the ACA and occurred across OECD countries with very different payment systems. The more plausible driver was a temporary lull in diffusion of expensive medical technologies, not payment reform.
Conceptually, VBP targets the wrong problem. U.S. costs are not mainly due to excess use of routine services; compared to peers, Americans often use fewer hospital days and doctor visits. The real drivers are
high unit prices,
adoption of costly new technologies, and
administrative overhead.
ACOs can’t control hospital pricing or tech adoption, and only capture a fraction of savings they generate.
Bottom line: VBP outsources cost control to intermediaries while leaving the core pricing and coverage decisions untouched—so it can’t solve long-term cost growth.
Header: More MACs Block Medicare Coverage of Advanced Imaging in Radiology.
##
I've often mentioned that experts interested in the growth of AI in pathology should look to advances in AI in radiology for signposts. See for example a 2022 blog on the topic here. (Also here). This may be even more important now, as AMA is advancing ideas originally thought up in the context of radiology, to apply to both radiology and pathology, which alarms me. 2026 blog here.
The CGS MAC brought out and finalizing, and the NGS MAC has just proposed, an LCD that doesn't pay for any kind of advancd diagnostic algorithms in MRI. Amazingly, the CGS MAC LCD got ZERO comments.
The LCDs are titled, "Automated Detection and Quantification of Brain MRIs."
The CGS MAC LCD DL40224 was proposed with comment period in October.
To my great surprise, it got no comments - A60362.
Now it's proposed newly for the NGS MAC region - as DL40332. Comment to March 8, 2026.
The gist of it is,
The use of AI augmented technology does not eliminate the need for appropriate reader training, and the results are not intended to be used independently of physician interpretation. National Government Services does not consider automated detection and quantification of brain MRI imaging to be reasonable and necessary.
Investigations have explored the potential of automated quantification technology for the evaluation of ARIA, MS, TBI, epilepsy, brain tumors, and other neurologic conditions. However, progress has been limited by the lack of established measurement standards and insufficient access to large, diverse datasets for training these tools....Without standardized data, it is difficult to determine whether subtle changes reflect true pathology or normal variation. At this time, there is insufficient evidence of clinical utility or validity, and use of this technology is considered investigational and therefore not covered. National Government Services will continue to monitor the evolving research related to these devices.
Continuing a saga that goes back 20 years, FDA goes after some companies not for their in-house LDTs, but for their collection kits shipped through the mails. (See a 2013 warning letter from FDA to 23andMe disallowing a non-approved saliva collection device.) (See a 2025 warning letter from FDA to DRG instruments that a saliva collection device was sold RUO, but with text referring to its "patient samples.")
According to a new item at 360DX, the FDA recently issued warning letters to:
Genovate
Germaphobix
ProDx Health
Genetrace. Online FDA warning letter January 2026 to GeneTrace here.
FDA's attention was especially high because direct to consumer marketing is involved. Actions could include "seizure, injunction, monetary penalties."
FDA noted that there are FDA-approved collection devices available (e.g. FDA protecting the market of companies that had invested in FDA approval.)
