Monday, September 22, 2025

Event: November 5: BLOODPAC "Therapy Selection Beyond DNA Mutations"

On November 5, 2025, BLOODPAC will hold a two-hour webinar on the interesting topic, "Therapy Section, Beyond DNA Mutations." RNA, digital pathology, multiomic dynamics.

Find more information and registration here.  1-3 ET, 10-12 PT.

https://www.eventbrite.com/e/new-frontiers-in-therapy-selection-beyond-dna-mutations-tickets-1708503553119

Writing,

In just a few weeks, BLOODPAC will host our fourth annual virtual New Frontiers seminar titled "New Frontiers in Therapy Selection: Beyond DNA Mutations." The event will take place on November 5th from 1:00-3:00 PM ET. Please use this link to register for the event.

This year’s New Frontiers Seminar will focus on the integration of multimodal data to inform and support therapy selection. The program will highlight emerging tools, such as RNA and epigenetic signatures, digital pathology, and more, and examine how they can be combined with liquid biopsy to capture real-time, multi-omic tumor dynamics. We’ll also discuss the challenges and opportunities of bringing these next-generation approaches into clinical practice to expand access and advance personalized cancer care.

Confirmed speakers and panelists currently include:

Speakers

  • Alexander Savchenko, Head of Diagnostic and Molecular Pathology, Novartis
  • Asaf Zviran, CEO & Co-Founder, PrismAI Therapeutics
  • Israel Tojal Da Silva, Head of Computational Biology and Bioinformatics, AC Camargo
  • Timothy Taxter, Executive Medical Director, AI Applications, Tempus AI
  • Tuvik Beker, CEO, Pangea Biomed

Panelists

  • Ben Glass, SVP, Product, AI and Translational Research, PathAI
  • Jim Godsey, Scientific Co-Chair, BLOODPAC (moderator)
  • Karim Azer, Global Head of Data and AI, Novartis
  • Matt Eaton, Head of Computational Biology and Bioinformatics, Precede Biosciences
  • Orly Alter, CSO & Co-Founder, PrismAI Therapeutics

###
AI CORNER
###
I gave the agenda to CHAT GPT 5 and asked it to expand on what the workshop MIGHT cover, in more detail.   Just for fun, here is what the AI guessed might occur.

###

 

AI TEXT FOLLOWS

BLOODPAC “New Frontiers in Therapy Selection: Beyond DNA Mutations”

November 5, 2025 | 1–3 PM ET (Virtual)

This year’s BLOODPAC New Frontiers Seminar promises a focused deep dive into how multi-omic and multimodal data are changing therapy selection. The two-hour agenda brings together leaders from pharma, diagnostics, and AI to discuss the move beyond DNA mutations toward integrated decision frameworks that capture real-time tumor biology. Here’s what this likely means in practice — and what to watch for.

1. Thematic Scope: Beyond DNA = Multi-Omic Therapy Selection

BLOODPAC’s framing (“beyond DNA mutations”) signals a pivot from purely mutational calls toward functional, dynamic, and multimodal signals:

  • cfRNA & Exosomal RNA: Measuring pathway activity, fusions, and splice variants that directly inform therapy selection.

  • Methylation Signatures: Sensitive at low tumor fractions, enabling tissue-of-origin calls and immune profiling.

  • Fragmentomics: Nucleosome and size patterns that infer chromatin state, tumor proliferation, and immune activity.

  • Proteomics & Tumor-Educated Platelets: Circulating proteins and platelet RNA as early-response indicators.

  • Digital Pathology: Whole-slide imaging (WSI) features like immune cell proximity, necrosis, or TLS density that complement liquid biopsy.

  • Composite Scores: AI-driven integration of these modalities into a single predictive index.

2. Speaker and Panel Contributions: What They Likely Add

Knowing the confirmed faculty lets us anticipate specific use cases and challenges that will surface:

  • Alexander Savchenko (Novartis, Molecular Pathology): Case studies on embedding RNA, epigenetics, and digital pathology into drug development pipelines. Likely to discuss regulatory-grade assay design and cross-site harmonization.

  • Asaf Zviran & Orly Alter (PrismAI Therapeutics): Computational biology innovators who will highlight how machine learning on RNA + methylation + liquid biopsy predicts response or resistance, including approaches for algorithm validation.

  • Israel Tojal da Silva (AC Camargo): Insights from a major cancer center on applying multi-omics in clinical trial cohorts and navigating real-world logistics (sample handling, assay reproducibility).

  • Timothy Taxter (Tempus AI): Experience with large-scale multimodal data fusion — probably a case where combining WSI + RNA + DNA outperforms single-omic models for patient stratification.

  • Tuvik Beker (Pangea Biomed): Perspective on pathway-level “network activity scores” derived from RNA, and how they can predict sensitivity to targeted therapies even in the absence of canonical driver mutations.

On the panel, expect:

  • Ben Glass (PathAI): Deep dive on using AI-powered histology features as a co-equal “omic” alongside liquid biopsy.

  • Karim Azer (Novartis): Pharma view on data integration at scale and challenges for regulatory acceptance and clinical trial endpoints.

  • Matt Eaton (Precede Biosciences): Practical considerations for deploying multi-omic assays in diverse patient populations, including reference ranges and normalization.

3. Likely Discussion Tracks

a) Therapy Selection When DNA Isn’t Enough

Examples where mutation calls are inconclusive — e.g., tumors with no actionable mutation, ambiguous VAFs, or pathway redundancy — and how RNA/methylation rescue the signal.

b) Real-Time Tumor Dynamics

Moving beyond baseline biomarkers: serial blood draws to track therapy response, clonal evolution, immune activation. Look for debate about sampling cadence and “how much change is enough” to alter treatment.

c) Digital Pathology + Blood Concordance

Integrating tissue and plasma signals:

  • Resolving false negatives in ctDNA by leveraging pathology-based tumor burden metrics.

  • Using WSI-derived immune contexture to decide if a patient is an I/O responder even when blood signals are borderline.

d) Analytics & AI

Managing heterogeneous, missing-by-design data:

  • Late vs. early data fusion.

  • Model drift and reproducibility in multi-omic pipelines.

  • Version control and regulatory reporting for AI-based algorithms.

e) Pre-Analytical Discipline

Blood tube chemistry, time-to-spin, RNA stabilization — critical for multi-omic reliability. Expect mention of BLOODPAC’s pre-analytical best practices playbooks and calls for community consensus on multi-omic QC metrics.

f) Regulatory & Reimbursement Pathways

  • Regulated unit question: Is the “test” the wet lab assay, the algorithm, or the combined system?

  • FDA change-control rules for composite tests — can you swap in a new RNA panel without a full re-clearance?

  • MolDx & CPT coding: Will multi-omic reports get one bundled code or modular coding (DNA + RNA + algorithm add-on)? This matters for pricing and adoption.

4. Clinical Utility & Trial Design

  • Designing trials that credit composite decision-making (RNA+DNA+IHC) rather than a single biomarker.

  • Using real-world data to justify coverage when randomized trials aren’t feasible.

  • Outcomes endpoints: replacing imaging scans with serial multi-omic MRD-style testing.

5. Consortium Deliverables

Expect BLOODPAC to preview:

  • Expanded multi-omic lexicon building on its ctDNA/MRD glossary.

  • Candidate reference materials for methylation and RNA assays.

  • Plans for shared datasets and benchmark challenges to accelerate algorithm validation.

6. Questions to Anticipate

  • “What is the clinical decision delta when adding RNA/methylation — beyond just higher AUC?”

  • “How do we validate a composite score when each component has different LoDs?”

  • “How do we manage regulatory submissions when AI models evolve?”

  • “How to price and code serial, multimodal tests for longitudinal care?”

Bottom Line

This seminar is poised to be less about abstract enthusiasm for multi-omics and more about how to make it work in practice — technically, operationally, and regulatorily. The lineup suggests robust case studies, frank discussions of assay standardization, and an emphasis on data fusion and deployment in real-world trials.

If you plan to attend, prepare to engage on clinical utility design, regulatory evidence frameworks, and practical multi-omic integration, as these will be front and center.

Would you like me to turn this into a one-page executive briefing or slide-ready deck (with key sections bolded and schematic diagrams for “multi-omic data fusion” and “clinical decision flow”)? That could make it easy to share with clients or colleagues ahead of the seminar.


[END AI TEXT]

Thursday, September 18, 2025

FDA Legally Rescinds "LDT Rule" Which Lost in Court

FDA has officially rescinded the LDT regulation which lost in federal court last spring.   The main act of recission was brief, merely removing a few words of text from the relevant regulation.   However, an appendix on financial costs and benefits annotated the billions of dollars of regulatory costs the move would have required, and the "billions" of dollars of "benefits" (largely speculative and cast in terms of dollarized life-years) the FDA had projected.

The regulatory change is signed by RFK Jr.

###


###

News items:

The rule was dead back around June 1, 2025, when a 60-day deadline to appeal the FDA's loss in court, was let pass.

One of my most-fun blogs of the past year was on December 25, 2024, when I fed the final legal arguments of FDA and its opponents into Chat GPT, which took the role of an "AI Judge" and predicted the FDA would lose.  Here.

###

Separately, FDA issued final guidance on enforcing LDT/IVD regulations during public health emergencies, "will balance risks and benefits."  Genomeweb here.  FDA PDF here.

###

###

AI CORNER

###

I asked Chat GPT to summarize the FDA's short publication rescinding the rule.  FDA makes no editorial comment about LDTs or future actions on LDTs.  But, FDA is obligated to publish a couple pages of changed financial projections.

###

FDA has formally rescinded its 2024 rule defining laboratory developed tests (LDTs) as medical devices after a federal court vacated the regulation in March 2025. The agency has reverted to the prior definition in 21 CFR 809.3, emphasizing that this action is ministerial and imposes no new compliance obligations. 

In its economic analysis, FDA highlights that vacating the rule saves an estimated $1.44–$1.54 billion annually (in 2024 dollars), mostly accruing to U.S. laboratories and suppliers, and is classified as a deregulatory action under Executive Order 14192. 

At the same time, FDA notes that the United States forgoes an estimated $3.7–$4.6 billion in annualized public health benefits, primarily from highly speculative “life-years gained” projections.

 The agency acknowledges that the uncertainty ranges for benefits and costs are wide enough to overlap, meaning the rule could have produced either net positive or net negative effects. FDA also observes that its quantitative models omit difficult-to-measure factors, such as potential impacts on innovation, underscoring the tentative nature of the original projections.

 


Tuesday, September 16, 2025

Digital Pathology Coding and Pricing: From Policy Questions to Policy Facts

Suspended Coding?

You'll be excused for thinking there are a lot of gray areas around digital pathology coding and pricing.  On the coding side, AMA CPT made about a half-dozen Proprietary Laboratory Analysis (PLA) codes for whole slide imaging (WSI) and advanced software (SaaS, AI, ML), then AMA seemed to declare a moratorium on new codes for similar services.  

New Coding?

Meanwhile, AMA CPT is holding feedback sessions on a possible new coding category called Clinically Meaningful Algorithmic Analyses (CMAA), which might encompass SaaS-like services for cardiology, radiology, pathology, and other fields.  Linked In here, AMA CPT here, here.

Pricing?

That's coding.  Meanwhile, CMS issued a call for comments on how it should price SaaS services (or AI, or ML) in Summer 2025 - here.

Some Solid Regulatory Touchpoints

This blog shows there is actually some solid ground here. Policy Facts follow.

#1 FDA Classifies Digital Pathology as "CLIA TEST"

First, with WSI and software based tests from Ibex (K241232) and ArteraAI (DEN240068), FDA classifies them first as PATHOLOGY tests, and then as CLIA HIGH COMPLEXITY TESTS.   This means they are CLIA tests for CMS purposes, and CMS regulations have pricing rules for the CLFS.   

See WSI and software classifed as CLIA High Complexity Test here (search keyword Ibex):


Note that FDA uses "PATHOLOGY" to encompass physician pathology (like a PMA CDx antibody) and clinical laboratory (like Illumina Trusight CGP which for FDA is "Pathology').



#2 CMS Defines Clinical Laboratory Test by using CLIA Text

To my knowledge, Medicare statute and regulations use the term "clinical laboratory test" and "clinical diagnostic laboratory test" (CLT, CDLT) but do not define them in an original way.   In regulations, this tends to me that a community common definition is clear enough.  For example, Medicare law doesn't define "appendectomy."  However, CMS does define "clinical laboratory test" in Benefit Manual, Chapter 15, 80.1.   There, CMS cuts and pastes language from CLIA law and uses it as a CMS definition of CLT.  

Section 1833 and 1861 of the Act provides for payment of clinical laboratory services under Medicare Part B. Clinical laboratory services involve the biological, microbiological, serological, chemical, immunohematological, hematological, biophysical, cytological, pathological, or other examination ofmaterials derived from the human body for the diagnosis, prevention, or treatment of a disease or assessment of a medical condition. Laboratory services must meet all applicable requirements of the Clinical Laboratory Improvement Amendments. (Ch 15:80.1).

The highlighted language is verbatim from CLIA (42 CFR 493).  CLIA uses this quotation to define a clinical laboratory, CMS uses the same language here to define a clinical test.  So if FDA classifies a digital pathology service as a CLIA test (Fact #1), that goes for CMS tos, per the definition above.

#3 Once a Service is a CLIA Test, It's Priced by CLFS Rules

Once a service is classifed as a CLIA test, it is priced by Clinical Laboratory Fee Schedule (CLFS) rules.  See SSA 1833, 1833A, 42 CFR 414.500ff.   The pricing mechanisms are crosswalk, gapfill, ADLT, or PAMA resetting.

The only exception to CLFS pricing of lab services is laboratory tests that CMS extracts and classifies as physician pathology services, which always require the personal sign out of a physician.   Based on 1970s-1990s law and ligitation, this is defined at 42 CFR 415.130.  (See also "AI Corner") at bottom.  As a reminder, FDA categorization of products uses "pathology" differently, not to distingish a physician service but inclusively meaning both pathology and clin lab tests.



#4 CMS Precedent.  WSI + SaaS = CLFS Test

CMS has clear precedent over several years of classifying PLA codes for WSI and software services as CLFS tests, the codes placed on the CLFS and the prices set by CLFS law and regulations.   Examples and precedents include 0108U, 0220U, 0261U, 0295U, 0376U, 0414U, 0418U, 0512U, 0513U.  (See also here.) 

We Can Conclude....

AMA CPT could put WSI-AI services into a new CMAA coding category, along with other SaaS like radiology services.  

However, FDA explicitly classifies WSI-AI services as pathology and CLIA services, from which point CMS has a defined pricing process (CLFS).  CMS also has PLA precedents which indicate the same.

##

##

##

AI Corner #1

See an AI version of the history of the debate over pathologists getting professional fees for each clin chem test; resulting law and regulations narrowing fees for a physician-pathologist lab service.  Here.

AI Corner #2

Summarize this blog.

Digital Pathology Coding and Pricing: AI Summary

Key Message:
The blog argues that despite the perception of "gray areas," there is a well-established regulatory and payment framework for digital pathology, whole slide imaging (WSI), and AI/SaaS-based pathology tests—both on the coding and pricing sides.

Coding: Current State and Future Directions

  • Existing Codes: AMA CPT has issued several PLA codes (e.g., 0108U, 0220U, 0261U, 0261U, 0295U, 0376U, 0414U, 0418U, 0512U, 0513U) covering WSI and advanced software services, which CMS has already placed on the CLFS.

  • Moratorium: After a handful of these codes, CPT paused creating new ones, indicating some hesitation about proliferating WSI/AI-specific codes.

  • Potential New Category: CPT is actively considering a new class of codes—Clinically Meaningful Algorithmic Analyses (CMAA)—that could standardize coding for SaaS/AI services across specialties (radiology, pathology, cardiology).

Pricing: CLFS Rules Apply

  • CMS Policy Clarity:

    • CMS explicitly asked for feedback in Summer 2025 on how to price SaaS/AI services, but its regulatory foundation is already in place:

    • Once FDA classifies a product as a CLIA test (as it did with Ibex WSI [K241232] and ArteraAI [DEN240068]), CMS must price it under CLFS rules (SSA §§1833, 1833A; 42 CFR 414.500ff).

    • Pricing methods are already well defined: crosswalk, gapfill, ADLT, or PAMA repricing.

  • Historical Precedent: CMS has consistently priced WSI/AI tests under the CLFS for several years, reinforcing that these services are treated as lab tests, not physician services.

Regulatory Foundations and Statutory Linkage

  • FDA’s Role: FDA classifies digital pathology products as pathology tests and CLIA high-complexity tests, putting them squarely in the lab regulatory space.

  • CMS’s Definition of CLT: CMS adopts CLIA’s definition of a "clinical laboratory test" verbatim in the Benefit Manual (Ch. 15, §80.1).

  • Physician Pathology Services Exception: The only services excluded from CLFS pricing are those narrowly defined in 42 CFR 415.130 as requiring physician sign-out (e.g., surgical pathology). Digital pathology software does not fall under this exception (per CMS decisions to date).

Conclusion

  • CPT’s Next Move: The AMA may group digital pathology and WSI-AI under the new CMAA category for coding consistency.

  • CMS’s Framework Is Ready: Since FDA classifies these services as CLIA tests and CMS has priced similar PLA codes via the CLFS, the regulatory and pricing pathway is already laid down. The challenge is not lack of policy but ensuring stakeholders recognize and use the existing pathway.



Monday, September 15, 2025

Tamara Syrek Jensen Leaves CMS, Joins Rubrum Advising

Tamara Syrek Jensen, an attorney who has had a long career at the Coverage Group at CMS, has moved to the consulting world.  

Syrek Jensen is now Principal and Vice President of Federal Programs at Rubrum Advising.   

Rubrum Advising was founded by Lee Fleisher MD, who was previously Chief Medical Officer at CMS and earlier, a professor at the University of Pennsylvania medical school.

###

Here's her bio at Rubrum.

https://www.rubrumadvising.com/

Tamara Syrek Jensen is a nationally recognized health policy leader and former Director of CMS’s Coverage and Analysis Group (CAG), where she shaped national coverage standards for devices, drugs, biologics, and preventive services supporting 65+ million Medicare beneficiaries.

As part of Rubrum’s work with government stakeholders, Syrek Jensen designs and executes commercialization strategies—bridging evidence planning, regulatory milestones, and payer engagement—from early development and pre-FDA authorization through full market launch.

Over more than 25 years at CMS, Tamara delivered major reforms that modernized coverage pathways and expanded access to care. She led creation of CMS’s new coverage pathway for FDA breakthrough devices—impacting more than 780 technologies and ensuring patients had timely access to innovations with proven outcomes. She modernized the agency’s Coverage with Evidence Development (CED) policy, cutting timelines for final guidance by more than half, and published landmark guidance documents that set new standards for transparency and predictability. Her work has been recognized in JAMA and NEJM and cited as a model for how CMS can drive innovation responsibly.

Before joining CAG, Tamara was Special Assistant to the CMS Chief Medical Officer and Director of the Office of Clinical Standards and Quality (OCSQ) (now CCSQ), where she was instrumental in developing the first MOU between CMS and FDA. That agreement laid the groundwork for the data modernization strategy that continues to support innovation today.

Tamara earned a reputation as a trusted leader and coalition-builder, fostering collaboration across CMS, FDA, NIH, specialty societies, and patient groups. Under her leadership, CMS doubled its evidentiary reviews, managed a $67 million budget, streamlined workflows, and consistently met all congressional timelines for coverage decisions.

She is admitted to the bar in Maryland and early in her career was a legislative assistant in the House.


Saturday, September 13, 2025

Geneoscopy Gets Reopening of NCD for Stool DNA Tests

A stool-DNA test from Geneoscopy, ColoSense, was FDA approved in May 2024 (PMA P230001).  

CMS covers the revised Cologuard test under the original 2014 NCD for Cologuard, but CMS couldn't stretch that NCD wide enough to include ColoSense.   On September 10, 2025, CMS officially reopened the NCD, which will now be "Screening for Colorectal Cancer - Non-invasive Biomarker Tests."  

The NCD 210.3 is currently titled "Colorectal cancer screening tests" and includes Cologuard, FOBT, and blood-based biomarker tests (screening LBx, Guardant).  Current document found here.

The time couse is slow:  Opening September 10, public comment September to October 10, and a proposed preliminary memo on March 10, 2026.   CMS anticipates completion of the final memo by June 8, 2026, which is 25 months after the FDA approval date of the product.

Find the home page for the policy revision here:

https://www.cms.gov/medicare-coverage-database/view/ncacal-tracking-sheet.aspx?ncaid=319

Read the Geneoscopy request letter here.  The request letter was submitted May 2024, about 16 months ago.  (And submitted a week after the FDA approval).

https://www.cms.gov/files/document/id319.pdf

The request letter runs 12 pages.

Geneoscopy raised $105M in growth capital in January 2025.  



##

AI CORNER

##

AI summary by Chat GPT 5.

TL;DR for Experts – Geneoscopy NCD Reconsideration Request (ColoSense™)

Clear Request:
Geneoscopy asks CMS to reopen and update NCD 210.3 (Colorectal Cancer Screening – Stool DNA Testing) to include stool RNA (sRNA) tests like its FDA-approved ColoSense™. They propose category-wide coverage language—not just for Cologuard—so any FDA-approved sRNA test with comparable sensitivity/specificity would be covered once every three years for average-risk beneficiaries aged 45–85.

Key Rationale:

  • Regulatory Trigger: ColoSense received FDA PMA approval on May 3, 2024 (with Breakthrough Device designation since 2020).

  • Clinical Validation: The pivotal CRC-PREVENT trial (n=8,920) showed 93–94% sensitivity for CRC (100% for Stage I) and performance metrics comparable to Cologuard’s 2014 data.

  • Public Health Need: Adds a second FDA-cleared, non-invasive, at-home stool-based CRC screen—critical to recovering missed screenings from COVID-19 and advancing Cancer Moonshot goals.

  • Policy Consistency: CMS has shifted toward category-based coverage (e.g., blood-based CRC screening tests in 2021, NGS in 2020); current sDNA-only language is outdated and forces one-off reconsiderations.

Proposed NCD Change:
Revise section 210.3.B.2 to explicitly include sRNA tests and set uniform coverage criteria (age, risk profile, frequency) mirroring those for Cologuard.

Bottom Line:
This is a well-structured, data-supported, and straightforward request for CMS to modernize NCD 210.3 by moving from single-test coverage (Cologuard) to category-wide coverage for FDA-approved stool-based sRNA tests, thereby expanding beneficiary access and aligning with [2024] CMS policy trends.

##

Sidebar - Chat GPT 5 rates the likely next steps. 


############################### footnotes

ColoSense got an updated collection device in July 2025.

##

Geneoscopy and Exact Sciences spent several years in patent litigation.

##

Generally, CMS covers screening tests under direct instructions from Congress or after endorsement by USPSTF.  However, statute allows CMS to create new screening benefits for colorectal cancer and prostate cancer by NCD, as occured for Cologuard in 2014 and as is occurring for ColoSense in 2026.

Friday, September 12, 2025

Oncomine Dx Express Test - FDA's 85-page Summary

In terms of transparency of its review process and standards, for genomic tests, FDA publishes a detailed "Summary of Safety and Effectiveness" within a few months of approving a new test kit.

FDA approved the Oncomine Dx Express Test on July 2, 2025.  The 85-page summary of the validation is online at FDA under PMA, P240040.  Home page here.

https://www.accessdata.fda.gov/cdrh_docs/pdf24/P240040B.pdf


The test has authorization for tumor mutation profiling in cancer patients with solid malignant neoplasms, as well as a PMA CDx authorization for Zegfrovy (sunvozertinib) with EGFR Exon 20 mutations.  The kit runs on the Ion Torrent™ Genexus Dx Integrated Sequencer.


##

Also available online at FDA is the summary of validation for Illumina Trusight Oncology Comprehensive.  Find under PMA P230011, from August 21, 2024. On NextSeq 550dx.  

Wednesday, September 10, 2025

Update on AMA Digital and AI Coding Frameworks: Tab 94: Thursday September 18

  • AMA may radically update its coding & concepts for AI, software.
  • Register to see the proposal AKA "Tab 94."
  • ALSO and separately, sign up to watch the AMA CPT meeting on September 18. 

###

AMA is considering very major changes to how it handles fundamental coding concepts for AI and software services.   (This comes at the same time that CMS is wrestling with payment and coverage issues).

See my recent blog on this AMA agenda item, called "Tab 94."

https://www.discoveriesinhealthpolicy.com/2025/08/digital-pathology-ai-and-coding-why.html

There, I provide instructions on how to register as a stakeholder interested in this topic and see the proposals to date, especially under the new possible category called "CMAA."

New News:  8 am Central on Thursday September 18, 2025

If you register (see above) to read the Tab 94 materials, that's one thing.

It's a separate workstream to register as a virtual participant in the September 18 AMA CPT meeting.  Register for that:

https://www.ama-assn.org/member-benefits/events/cpt-editorial-panel-meeting

AMA will track you by email and update you on the schedule.   Right now, AMA sees the Tab 94 discussion as 8 am Central Time on Thursday September 18.  But register to track updates.


Genetic CPT Codes in 2024: The Massive Boom in Epilepsy Code 81419 (NOVITAS Again!)

This past week, CMS released data on lab test utilization by CPT code for CY2024 and CY2023.  I published a rapid blog on this and looked at the distribution of payments by proprietary PLA code.

https://www.discoveriesinhealthpolicy.com/2025/09/cms-releases-part-b-utilization-by-cpt.html

I've gone back and looked at the CPT code series 811nn, 812nn, 813nn, 814nn - these are the non-proprietary genetic CPT codes.   Here are the top codes paid in 2024. Click to enlarge

click to enlarge, top codes in series 811 812 813 814

The Top Code is 81479

This category of code was paid $1.4B by CMS Part B in 2024, of which 42% or $584M went to Unlisted Code 81479.   Historically, this is paid 90% or more only in MolDx states.

What The F*** With 81419, 81440, 81443???

Most of the other top codes are from the roughly dozen codes now assigned to comprehensive genomic profiling (code 81445 forward, CGP codes).

However, it's odd to see Epilepsy Panel 81419 as the second-highest code, paying $74M.   

This code grew massively in 2022 and 2023 and 2024, just as notorious code 81408 did in 2018-2021 (reaching about a billion dollars paid.)   For 81408, see OIG report $888M lost.  Blog 2023 on 2021.  Blog 2021 on 2019.

In 2022 there were just 41 payments for 81419, epilepsy panel, in Part B.


For 2023, where we have data by lab and by state, we see the 2023 payments were entirely in Novitas MAC states and FCSO (Florida).    This is about the same pattern seen earlier in 2018-2020 for 81408.

See column of States at left = Novitas & FCSO MACs

Billing 81419 was Concentrated

About 57% of national Medicare payments for epilepsy panel in 2023 went to just 4 Novitas-FCSO labs (in FL, NJ, TX).

Labs Billing 81419 Were Atypically of LLC Type

Of 27 labs that billed 81419 in 2023, 16/27 or 60% were "LLC."  (Of all labs billing popular code TSH 81443, 27% were LLC (301/1104).   See AI comments about the actual lab names below.

Labs Billing 81419 Often Co-Billed 81440, 81443

In 2023, of labs billing any of these unexpected genetic codes (81419, 81440, 81443), 10 billed all three codes (10/27, 37%).  

This group of codes (81419, 81440, 81443) was also paid by Medicare [essentially] in only Novitas and FCSO jurisdictions.  81440 is mitochondrial panel >100 genes, $3224, 81443 is inherited severe conditions like Ashkenazi panels, $2449. The total damage?  Almost $200M.   Click to enlarge

click to enlarge


99% of Payments "NOT" to GeneDx, Ambry, Quest, Labcorp, etc.

In 2023, there were $27M in payments for (81419, 81440, 81443), for 10,832 Medicare patients, of which 123 cases were for familiar labs GeneDx and Labcorp (1.1%).

Conclusion

The program vulnerability at Novitas-FCSO was unchanged from 2018 to 2024, allowing rare genes to explosively rise to the top of the payment cycle with no intervention.   The same behavior was not allowed at any other MACS - not MolDX MACs or NGS MAC.   Despite repeated statements (up to today) from CMS about interventions with super duper AI against fraud, the situation with 81408 and similar codes can be picked up by a nine year old with an Excel spreadsheet. 

###

AI CORNER

###

In labs billing 81419, I didn't see major brands like Ambry and Quest.   I asked Chat GPT to read and discuss a couple dozen lab names that were billing 81419 in 2023.  

  • Prestige or Elite Branding
    • Use of terms like "Landmark" or "Fidelity"
  • Science-y
    • Gene Express, Thesis Genetics
  • Obscure
    • Pro Exonics LLC
    • Genomic Pathology UltraSeq LLC
  • Odd Shell Names
    • ExelproDx Labs LLC
  • None of the 81419 labs mention a state or locality.  Formulaic names were reused (Exon Express Genetics).  LLC name was common.

##

(For more on odd lab names here).

##

Chat GPT  Recognizes Anomalous Names

I gave 3 paired short lists of lab names to ChatGPT 5 and asked whether, in each pair, it could pick the list of anomalous or unusual labs.  It was correct 100% of the time.

NPIs Often Brand New

I spot checked a few NPIs and they were all from 2022 or 2023 (for labs paid millions in 2023), and some were out of business by 2025.

Multi-Code Billing Per CMS Patient

Another flag was, if ABC Lab had 1,206 Medicare cases of 81419 Epilepsy gene panel, it also happened to have 1,206 Medicare cases of 81440 (Ashkenazi panel) and 1,206 Medicare cases of 81443 (mitochondrial gene panel).  This parity among the 3 codes was never seen at Ambry, Quest, GeneDx, or LabCorp.  It reminds of the 81408 situation of 2020, where a lab would bill for 1,206 cases of rare gene 81408, also for exactly 1,206 cases of rare gene 81407, of 81406, of 81405.

##

AI-Generated Monolog

See an AI generated complaint in the voice of comedian Lewis Black about the Medicare payments - here.

##

Novitas Controls?

Novitas and FCSO have a generic article that lists hundreds of gene code and makes a generic statement that services must be reasonable and necessary - A58918.

##

FFS vs Medicare Advantage??

Another quick scan for CMS detection of "anomalous claims," would be pull up labs that ONLY bill genetics to FFS Medicare patients, and never to the majority of Medicare patients who are in Medicare Advantage.  I can't run that query.

##

These Lab Over At MolDX DEX Registry

Of 27 labs paid for 81419 in 2023, aside from GTI AZ and GENEDX MD, 2 labs were listed and had genes on offer but none showed  approved or payable from MolDx.  3 others were listed in DEX registryby lab name only but no tests had been submitted.

Tuesday, September 9, 2025

CMS Releases Proposed Lab Code Crosswalks & Final Gapfill Pricing

On September 9, 2025, CMS released proposed new lab test pricing for 2026, for public comment.

CMS also released final Gapfill pricing for 2026. 

Go to the CMS annual public meeting lab website here:

https://www.cms.gov/medicare/payment/fee-schedules/clinical-laboratory-fee-schedule-clfs/annual-public-meetings

Scroll down to Meeting Notice & Other Important Materials.

Under that, scroll to find CY2026 Preliminary Payment Determinations, and to find 2025 Final MAC Determinations.

###

GAPFILL - FINAL

There were 34 codes under Gapfill. 
Six codes for neurologic tests (amyloid, tau) rose from $70-93 to either $116 or $128.

Nine other codes raised in price got 130%, 211%, 125%, 244%, 1234% (!, 440U) 166%, 166%, 125%, 131%,     440U was a plasma metagenomic test that rose from $36 to $444, more than 10X.

In all, about half the codes (6+9=15, of 34) had price increases based on supplemental information provided over the summer.

###

NEW CODES - PRELIMINARY

These are my manual tallies, so I could be off by one or two.

There were 90 codes in play.   In 57, CMS agreed with the expert panel majority recommendation.  In 34, they did not.   Of those 34, only one agreed with a "substantial" minority vote of 4.   Otherwise, CMS either disagreed with all the panelists or all but 1 of the panelists.  Quite a few of these 34 were cases where CMS chose a proposal that no panelist had voted for and in some cases, CMS chose a proposal that had not even been mentioned in discussions.

I counted 21 of 90 prelminary proposals being for 2026 "Gapfill."

There was one case - 0523U, Item 26 - where CMS specifically chose an FDA-approved crosswalk for a new FDA-approved test.  Pillar Oncoreveal CDx (22 genes).   It is crosswalked to "0022U 81449".  0022U is $1950.   81449 is $597.   CMS writes this as "0022U - 81449" with a dash or minus sign rather than a plus sign.  

##

BONUS - CMS Dislikes Fracitional Crosswalks

##

At a talk on October 23, 2025, Erik Schulwolf of Hogan law firm had this quote from CMS on a slide:

CMS is permitted under the regulations to crosswalk to a fraction or multiple of a test on the CLFS
– Agency has in the past stated that “in an effort to maintain consistency across the entirety of recommendations, in most cases we support and recommend to crosswalk the new codes without multipliers”

I took a photo of the slide, uploaded to Chat GPT for a "transcript."   I then plugged the quotation into Google, which gave me a "hit" for the quotation source - the CMS 2019 CLFS pricing decisions, made in about November 2018.

https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/ClinicalLabFeeSched/Downloads/CY2019-CLFS-Codes-Prelim-Determinations.pdf




MolDX New Cancer LCD: For Non-NGS Testing in Cancer

On August 28, 2025, I flagged two new proposed LCDs from MolDx, one for thrombophilia (F5) testing, one for liver disease steatosis risk.  

MolDX released a third proposed LCD on August 28, titled, "Non-Next Generation Sequencing Targeted Molecular Panel Tests for Predictive Testing in Cancer."  

Find it here, DL40210.  There is a 2/2024 request letter from Padma Sundar and a 6/2024 request letter from Anjana Bhattacharya (Biofidelity).  The LCD has 81 citations.  Comment til October 12.

The LCD has a draft billing article DA60236 here.  (The only code listed is 81479 plus DEX Z-code.)

  • Note that this is only relevant for outpatient biopsies.   If the patient has an inpatient colectomy, for example, you're free to order PCR on Day 1 and NGS on Day 4 but both are bundled (nonpayable) under the 14 day rule for inpatients.
  • See an article on Biocartis IDYLLA platform and rapid targeted CDX as a business plan.  Here.

The Rules

MolDx opens the LCD 9 rules, some of which are general ("the test has been validated," "the patient has cancer," etc).   Specifically, the non-NGS test can be used when at least one of these is true: (1) NGS is not feasable or likely to fail, OR (2) NGS is not required by guidelines and non-NGS testing will provide management, OR (3) the test includes necessary biomarkers and precludes unnecessary NGS testing.  A negative result in a non-NGS test may be followed with a larger NGS test.   The non-NGS test includes the relevant genes for FDA-approved therapies.

  • For me, it's tempting to read the 3 restrictions as "AND" - making the gateway to PCR testing very narrow - but it's clearly written as "OR" - making the gateway pretty wide.

Genomeweb 360Dx reports the LCD is relevant for ChromaCode HDPCR, Biofidelity Aspyre Lung, and Biocartis Idylla [platform] assays.

###

AI CORNER

###

Playfully, I first gave Chat GPT 5 only the bibliography of the LCD, and asked it to guess what the body of the LCD (not shown) is about.

###

Based on the bibliography you shared, this new MolDx draft LCD is focused on PCR-based panel testing in solid tumors (especially NSCLC and colorectal cancer) — not on broad NGS panels.

Here’s why:

1. Dominant Focus on NSCLC

  • Many references are about EGFR, ALK, ROS1, RET, MET exon 14, KRAS, BRAF and other actionable drivers in non-small cell lung cancer (e.g., Kris 2014 JAMA; Lindeman/CAP-IASLC-AMP guidelines; multiple NCCN guideline versions; numerous papers on turnaround time and reflex testing).

  • There are repeated mentions of reflex single-gene or small multiplex assays and their impact on time-to-treatment and outcomes.

2. Emphasis on PCR Platforms

  • Several papers specifically evaluate Idylla (Biocartis), ASPYRE, digital PCR, and rapid cartridge-based systems (Evrard 2019, Momeni-Boroujeni 2021, Banyi 2022, Carnero-Gregorio 2025, Buglioni 2022, Depoilly 2022, Melchior 2024, Hogarth 2025).

  • These are PCR-based methods for detecting actionable variants and fusions, positioned as alternatives or complements to NGS.

3. Secondary Focus on Colorectal Cancer

  • References to KRAS, NRAS, BRAF, and PIK3CA mutations in metastatic colorectal cancer (De Roock 2010; Therkildsen 2014; Al-Turkmani 2018).

  • Again, much of the cited literature is about PCR-based testing rather than large CGP panels.

4. Themes in the Citations

  • Speed and turnaround time: rapid results enabling earlier treatment decisions.

  • Clinical utility and cost-effectiveness of rapid PCR panels.

  • Comparisons of PCR vs. NGS (sensitivity, sample requirements, concordance).

  • Workflow integration in pathology labs (reflex testing, small FFPE samples).

✅ Interpretation:
This draft LCD (DL40210) is not about broad NGS panel coverage, but rather about coverage for rapid, multiplex PCR-based panel testing in cancer tissue — with the main clinical use cases being NSCLC driver mutations/fusions and RAS/RAF testing in colorectal cancer.

It looks like MolDx is carving out a policy lane for automated cartridge-based PCR panels (e.g., Idylla, ASPYRE, ChromaCode, etc.) as an alternative to broad CGP, emphasizing faster turnaround and use in tissue-limited scenarios.

 ####

Now I gave Chat GPT the whole LCD to discuss.

###

Here’s a structured take for your genomics industry blog readers on MolDx’s new draft LCD DL40210: Non-NGS Targeted Molecular Panel Tests for Predictive Testing in Cancer:

1. Criteria for Coverage vs. Non-Coverage

MolDx lays out a detailed set of conditions for when non-NGS (PCR or similar) multigene cancer panels are covered:

  • Patient eligibility:
    Must have a cancer diagnosis where molecular results will inform treatment.

  • When NGS is not the first choice: Coverage applies if:

    • NGS is not feasible (poor specimen, low tumor content).

    • NGS isn’t required by guidelines, and a faster non-NGS method can guide timely treatment.

    • The non-NGS test can safely identify common actionable biomarkers and help avoid unnecessary NGS.

  • Testing rules:

    • No duplicate testing of the same content; negative results may be reflexed to NGS for additional targets.

    • Panels must cover the most common guideline-recommended biomarkers (e.g., EGFR, ALK, ROS1, RET, BRAF, METex14, NTRK, KRAS/NRAS in CRC).

    • Test must be validated for the intended population and sample type.

    • Performance must be comparable to NGS, with faster turnaround (<10 business days, ideally 1–3 days).

    • Must pass MolDx technical assessment (analytical validity, clinical validity, clinical utility).

  • Non-coverage (implied):

    • Panels that duplicate NGS without a turnaround/time-to-treatment benefit.

    • Tests not validated, with poor concordance to NGS, or that miss common guideline biomarkers.

    • Platforms that only cover rare targets or provide incomplete content for standard-of-care therapy selection.

2. What MolDx’s Goals and Concerns Likely Are

Although MolDx (and CMS via NCD 90.2) strongly supports broad NGS panels, this LCD signals pragmatic flexibility. Likely goals:

  • Address Turnaround Delays:
    Long waits for NGS results (2–4 weeks) often push oncologists to start empiric chemo or immunotherapy, risking worse outcomes. Rapid PCR-based panels (Idylla, ASPYRE, ChromaCode, etc.) can deliver results in hours to days.

  • Tissue Constraints:
    Small biopsies or cytology samples often fail NGS. Cartridge-based PCR panels can run on low-input or degraded samples, rescuing otherwise “quantity-not-sufficient” cases.

  • Encourage Tiered Testing Models:
    Allow labs/clinicians to use fast PCR panels for the high-frequency drivers, then reflex to NGS if negative. This improves efficiency, tissue stewardship, and may reduce unnecessary send-out NGS.

  • Balance Innovation with Guardrails:
    MolDx insists on peer-reviewed validation, high concordance, and low false-positive rates. They don’t want rapid panels to become a loophole that undermines the rigor of CGP or leads to inappropriate therapy.

  • Cost and Utilization Control:
    Coverage is framed to prevent duplicative billing (PCR panel and full CGP for every patient). The LCD stresses that multiple assays may constitute “one service,” signaling attention to economic efficiency.

3. Big Picture Takeaway

This LCD is not a retreat from NGS CGP—MolDx and CMS continue to endorse large panels for comprehensive profiling. Instead, it’s a practical bridge policy: recognizing the clinical value of rapid, non-NGS panels in situations where speed or tissue constraints trump breadth.

Think of it as CMS acknowledging:

  • NGS = gold standard, but not always practical.

  • PCR = fast, pragmatic, and valuable in the right context.

  • Together, they form a tiered ecosystem of tumor testing, with PCR as the “first strike” in some patients and NGS as the “catch-all” for negatives and rarer events.

 

###

How should major NGS test providers view this LCD? How would biopharma view it? It sounds like most of the tests are non FDA, but i might be wrong on that. How does this fit with some centers like MGH, that have advocated ultra rapid early panels (like EGFR in Lung Cancer) followed in 7-14 days by NGS panels.

###

Great set of questions. Here’s how I’d unpack this for your genomics strategy blog audience:

1. How Major NGS Test Providers Should View This LCD

For the large comprehensive genomic profiling (CGP) companies (Foundation, Caris, Tempus, Guardant, etc.), this LCD is both complementary and slightly competitive:

  • Complementary:
    MolDx still affirms that NGS is the gold standard for comprehensive testing. The LCD explicitly allows non-NGS panels as a triage step, with reflex to NGS if PCR is negative. This reinforces the role of CGP in covering rare variants, resistance mutations, tumor mutational burden, MSI, and tumor-agnostic biomarkers that PCR cannot detect.

  • Competitive Risk:
    If rapid PCR panels proliferate, oncologists may be more inclined to treat based on those results alone, delaying or skipping CGP in some cases. That could shrink the immediate clinical “pull” for NGS—especially in community oncology, where the turnaround problem is most acute.

  • Strategic Opportunity:
    NGS providers may benefit by positioning themselves as the second step in a tiered pathway: e.g., “we’re the comprehensive follow-up to the rapid PCR triage.” They may also co-market with rapid PCR vendors or even acquire/licence those platforms to offer bundled solutions.

In short: NGS providers should see this as a policy endorsement of a hybrid model rather than a threat to CGP’s long-term centrality.

2. How Biopharma Should View It

For drug developers, the LCD is actually a win:

  • Faster Patient Identification: Rapid PCR panels help identify patients with common, druggable mutations (EGFR, ALK, KRAS, BRAF, etc.) much earlier. This means more patients can get on targeted therapies (often in first line) instead of missing the therapeutic window.

  • Supports Companion Diagnostic Uptake: Even if many of these PCR panels aren’t FDA-cleared CDx devices, MolDx coverage makes it easier for oncologists to order guideline-mandated biomarker testing and match patients to approved drugs.

  • Broader Access in Community Settings: Biopharma is well aware that delays and tissue inadequacy in NGS testing are barriers to adoption. This LCD lowers those barriers, particularly outside academic centers.

The caveat: most of the cited tests (Idylla, ASPYRE, ChromaCode HDPCR, etc.) are not FDA-cleared CDx assays but are LDTs running under CLIA. MolDx is effectively giving these non-FDA platforms a reimbursement pathway, as long as they pass MolDx’s technical assessment. That’s unusual but consistent with MolDx’s long history of supporting LDTs where commercial CDx tests don’t fully meet clinical needs.

3. Fit with Ultra-Rapid Panel Strategies (e.g. MGH)

Centers like Mass General/Harvard have long advocated a two-step strategy:

  • Step 1: Run a very rapid single-gene or small-panel PCR test (e.g. EGFR mutations in NSCLC), with results in 24–72 hours. This avoids empiric chemo/IO in patients who might instead benefit from immediate targeted therapy.

  • Step 2: Run full NGS (7–14 days turnaround), to capture rarer drivers, resistance mutations, and trial-eligible markers.

This LCD institutionalizes that philosophy. It validates that rapid PCR testing is clinically valuable in the early window but insists that negative or incomplete cases should reflex to NGS. In other words, MolDx has effectively translated the MGH playbook into reimbursement policy.

4. Bottom Line

  • For NGS companies: This isn’t a rollback; it’s a codified two-tier model. But they must adapt messaging: “NGS is the essential complement to rapid PCR triage.”

  • For biopharma: It helps ensure more patients get on their targeted therapies quickly—boosting adoption curves.

  • On regulatory status: Most of these rapid PCR panels are non-FDA, CLIA LDTs, but MolDx’s willingness to reimburse them (subject to technical validation) means the FDA gap is not a barrier in the Medicare population.

  • In practice: This LCD brings Medicare policy into alignment with what leading cancer centers (like MGH, MSK, MD Anderson) already do—rapid initial panel + comprehensive NGS.

Would you like me to also sketch how private payers are likely to interpret this? (Some may be less inclined than MolDx to pay for both PCR and NGS, unless carefully staged.) That might round out the blog post with a payer-market lens.

##

Perfect — let’s round this out with a payer-market lens, since your readership is always alert to what happens after MolDx moves first.

1. Private Payers Will Watch Closely

Commercial insurers generally take their cues from MolDx and CMS, but with more caution on cost duplication. They’ll note two things immediately:

  • MolDx frames rapid PCR panels as reasonable and necessary only when they prevent delay or specimen failure.

  • MolDx emphasizes that multiple assays may constitute a single service — i.e., don’t expect to bill PCR panels and NGS universally for every patient.

Private payers are likely to echo this emphasis: coverage is for targeted use, not carte blanche.

2. Expect Guardrails on PCR + NGS “Double Dipping”

Commercial payers are already wary of NGS utilization growth. From their perspective:

  • PCR panel alone: Acceptable if it replaces NGS for patients with limited tissue or acute need for fast results.

  • PCR followed by NGS reflex: Acceptable if the PCR is negative or incomplete.

  • PCR + NGS in every patient: High risk of denial. Insurers will see this as duplicative and not aligned with MolDx’s intent.

So expect prior authorization requirements or medical policy language requiring documentation of why both were needed.

3. Alignment with Broader Payer Trends

  • Time-to-treatment: Payers are sensitive to oncology outcomes tied to fast biomarker results. This LCD gives them language to support “early actionable biomarker detection” while still steering patients to CGP when needed.

  • Site of care dynamics: Community oncology groups (often under value-based care contracts) may find PCR panels attractive because they’re fast, low-tissue, and potentially cheaper than NGS send-outs. Private payers will favor models that reduce unnecessary NGS but don’t miss drug-eligible patients.

  • FDA status: Unlike Medicare/MolDx, commercial payers may raise more questions about LDT vs. FDA-cleared CDx. Some already write policies preferring FDA-cleared companion diagnostics when available. They could therefore be slower to embrace some of the PCR platforms cited in the LCD (Idylla, ASPYRE, ChromaCode), unless academic validation is strong.

4. Strategic Implications

  • NGS labs: Need to educate payers that PCR-first pathways risk under-detection of rarer drivers and trial markers. Their argument: “comprehensive NGS is the definitive test, PCR is an optional triage.”

  • PCR vendors: Will push the cost-savings angle (avoid unnecessary NGS, faster therapy initiation, fewer empiric chemo/IO starts).

  • Biopharma: May advocate with payers that rapid PCR triage improves patient capture for on-label targeted therapy, reducing waste on ineffective treatments.

5. Bottom Line for Private Payers

Private insurers are likely to follow MolDx’s tiered testing framework but with tighter utilization controls. Look for policies that:

  • Require medical necessity documentation for PCR + NGS.

  • Prefer FDA-cleared tests when available.

  • Restrict broad coverage to cancers where guideline biomarkers are clear (NSCLC, CRC, melanoma, thyroid).

For the industry, this means PCR-NGS tiering is now legitimized by Medicare, but private payers will push hard to ensure it’s “either/or + reflex,” not “both, always.”

 ###

 300 WORD SUMMARY

###

Here’s a concise 300-word blog-ready summary:

MolDx’s new draft LCD DL40210 carves out a role for non-NGS (PCR or similar) multigene panels in cancer testing. Coverage is allowed when patients have cancer where results inform therapy, and when NGS is either infeasible (low tissue, poor quality), not required by guidelines, or unnecessarily delays care. Panels must cover the most common actionable biomarkers (EGFR, ALK, ROS1, RET, BRAF, METex14, NTRK, KRAS/NRAS in CRC), be validated in the intended setting, deliver results faster than NGS (<10 business days, ideally hours–days), and pass MolDx technical assessment. Non-coverage applies to tests that duplicate NGS without a turnaround benefit or miss guideline-mandated targets.

MolDx’s goals are clear: reduce delays tied to NGS send-outs, enable testing of low-yield tissue, and support tiered strategies where rapid PCR identifies high-frequency drivers, with reflex to NGS if negative. Guardrails are emphasized: avoid duplicative billing, demand strong validation, and ensure positive PCR results can directly guide therapy.

For NGS companies, this is both complementary and competitive. It reinforces CGP as the “gold standard” for rare drivers and tumor-agnostic biomarkers, but PCR triage could siphon some early treatment decisions. Strategically, NGS providers may reposition themselves as the indispensable second step in a hybrid pathway or partner with rapid PCR vendors.

For biopharma, the LCD is a win. Faster PCR identification of common drivers means more patients start on targeted therapies in first line. Even though most of these assays (Idylla, ASPYRE, ChromaCode) are non-FDA LDTs, MolDx reimbursement lowers barriers to adoption, particularly in the community setting.

Private payers will likely follow MolDx but with tighter controls: coverage for PCR alone or PCR-then-NGS reflex, not “PCR plus NGS for all.” Expect prior authorization and preference for FDA-cleared tests.

In short, MolDx is institutionalizing the MGH-style hybrid model: rapid PCR upfront, NGS for breadth—bringing policy in line with what leading cancer centers already do.

###

###

Obviously, we can't read the mind of MOLDx, but does it seem they have a negative stance - wanting to avoid paying for both upfront PCR and later NGS - or a positive stance - wanting to be sure that patients in key scenarios can get rapid focused testing while perhaps awaiting larger panels that might, for example, support TMB.

###

You’re right — we can’t know MolDx’s inner stance, but the tone of DL40210 is more positive than negative. A few signals stand out:

  • Not written as a crackdown: The LCD doesn’t frame PCR panels as wasteful or duplicative. Instead, it goes to great lengths to describe situations where PCR panels are reasonable and necessary (small specimens, urgent treatment needs, avoiding empiric chemo/IO).

  • Explicit emphasis on timeliness: They cite evidence that long NGS turnaround drives inferior outcomes, and that rapid PCR results can materially improve survival. That’s a clear endorsement of PCR’s role, not a warning shot against it.

  • Guardrails, not prohibitions: The language about multiple assays counting as “a single service” and reflex-to-NGS only after negative PCR is less about discouraging PCR, and more about preventing systematic double billing. It’s fiscal discipline layered on top of a clinically positive message.

  • Consistency with practice at leading centers: By echoing the “MGH model” (rapid PCR triage + NGS follow-up), MolDx is signaling they want Medicare policy to align with best-practice workflows, not fight them.

  • Still affirming NGS as the gold standard: They don’t minimize the importance of CGP for rare alterations or tumor-agnostic markers. The policy makes PCR panels a bridge, not a substitute.

Bottom line

The stance looks pragmatically positive: MolDx is not trying to wall off NGS or punish labs for running PCR first. Rather, they want to legitimize PCR panels in key scenarios (speed, tissue limits) while ensuring they don’t become a redundant add-on to NGS. The goal is patient-centered—get results fast when it matters most—while keeping utilization and costs in check.

####

It seems potentially confusing since if I am offering an NGS test, I would look to the NGS CGP LCD for coverage rules. But I'd also have to unexpectedly look to this different PCR LCD, which has some possible restrictions on NGS CGP even though the title and topic is all about non NGS! Confusing.

###

Exactly — you’ve spotted one of the more confusing features here.

On its face, DL40210 is titled as a policy for non-NGS tests, but buried inside is guidance that directly affects NGS coverage in practice. A lab offering an NGS CGP panel would naturally look to the long-standing MolDx NGS LCDs (L38045, L38047). But in reality, DL40210 sets up rules about when PCR panels should be used instead of NGS, or when PCR results should precede NGS reflex. That creates an overlap zone:

  • From the lab’s perspective: You’re required to comply with both the NGS LCD and this new PCR LCD. If you ignore DL40210, MolDx could question why you billed NGS up front when the LCD describes scenarios where PCR triage is “reasonable and necessary.”

  • From the oncologist’s perspective: It muddies the waters. The message has always been “NGS is endorsed by CMS.” Now the LCD inserts a carve-out for rapid PCR — not as a competitor, but as a preferred first step in some patients.

  • From a policy perspective: The title makes it sound like a niche non-NGS policy, but the text effectively conditions NGS coverage on considering PCR first in certain scenarios. That’s what feels unexpected.

Why this matters

For NGS companies, the risk isn’t that MolDx is retracting support — it isn’t. But it does mean payers could interpret DL40210 as a utilization management tool, nudging clinicians away from “NGS for everyone on day one” and toward a tiered model. That could complicate billing, documentation, and pre-authorization.

In short: yes, it’s confusing. A policy ostensibly about non-NGS tests ends up shaping the context for NGS CGP. Labs and oncologists will now need to read these LCDs together, not separately.

###

FDA VS RUO

I don't think many of these PCR tests are fully FDA approved - but I could be wrong.  Assuming that is correct (and I'd happy to revise it), would these "kits" need to be labeled RUO in the USA?

FDA & IDYLLA see an older 510K for MSI testing for Lynch genetics, and a new 2025 PMA for MSI as a CDx.

###

You’re right to flag that distinction — it’s a subtle but important one.

1. FDA Approval Status

Looking at the assays referenced in the bibliography and the LCD (Idylla™ EGFR, Idylla GeneFusion, ASPYRE, ChromaCode HDPCR, digital PCR panels, etc.):

  • FEW are FDA-cleared companion diagnostics (with the exception of the Oncomine Dx Target Test FN43, which is NGS-based and FDA-approved, but that’s not the focus of this LCD; and Idylla "MSI").

  • The rest are laboratory-developed tests (LDTs) running on commercial hardware platforms (e.g., Biocartis Idylla, ASPYRE cartridge-based systems). These are manufactured kits, but they are not cleared or approved as IVDs by FDA.

So your instinct is correct: they are not FDA kits in the CDx sense.

2. Regulatory Labeling

Because they are distributed as kits, but without FDA clearance, they must be labeled “For Research Use Only. Not for use in diagnostic procedures” (RUO) under FDA rules. That creates a paradox:

  • The physical cartridges and platforms are marketed commercially, often with CE-IVD labeling in Europe.

  • In the U.S., absent FDA approval, they are RUO kits — even though in practice they are run in CLIA labs as part of validated LDTs.

3. MolDx’s Position

What MolDx is doing here is essentially saying: “We don’t require FDA approval for these panels. If you validate them as LDTs under CLIA, and if they pass MolDx technical assessment, we will cover them.”

This is consistent with MolDx history: they’ve long reimbursed non-FDA NGS LDTs (FoundationOne before FDA approval, Tempus, Caris, etc.). The difference here is they’re extending the same policy flexibility to rapid PCR kits running as LDTs.

✅ Takeaway:
These are indeed kits running on manufactured platforms, but since they’re not FDA-cleared, they must be labeled RUO. MolDx is effectively stepping in to provide reimbursement despite the RUO labeling, relying on its own technical assessment as the arbiter of clinical validity and utility.

 

 If you enjoyed this LCD on the dynamics of PCR testing, you might enjoy:

Liu L et al. (2025) Research Trends and Development Dynamics of qPCR-based Biomarkers: A Comprehensive Bibliometric Analysis.  Molecular Biotechnology (1/22/2025).  PMID 39843617.  

Can Chat GPT Psychoanalyze MolDX?

Can CHAT GPT 5 psychoanalyze MolDx?  We feed it some documents to find out.

###

One of the flagship uses of AI is to summarize things - to summarize 3 PDFs as mini-review of the 3 peer-reviewed papers, to summarize a long New York Times article, and so on.    

But here's a more offbeat use:  Can Chat GPT make sensible inferences about what the human author(s) of a document was really thinking?

This month, MolDX held a contractor advisory meeting (CAC) on the topic of molecular tests to classify Barrett's esophagus.    While the transcript hasn't been released yet, MolDX did post a list of about 15 questions for the expert panelists.   I uploaded that one item to Chat GPT and asked for feedback, if any, on what the MolDx author(s) were thinking.   It found the questions were written in a balanced way, yet some denote a tone of skepticism - and so on.

All well and good.  Then I noticed that MolDx had also held a CAC on this topic in 2021, and had also written a noncoverag LCD in this area in 2023.  

That gave me a new idea.  This time, I took all three documents - questions 2021, LCD 2024, questions 2025 - and gave them to Chat GPT and asked for an assessment of the "evolution" of MolDX thinking, and if this longitudinal view gives more insights into the current 2025 questions.   Chat GPT gave a detailed assessment and found that it did.

###

Not Right or Wrong

This is an area without a clear "right or wrong."  It's more like, if you ask 10 college students to review The Great Gatsby, you'll get ten different essays.  Some will like it, some will hate it, some will focus on characters, others on plot, etc.   Similarly, there 's not right or wrong way to look at the three MolDx PDFs and discuss what the authors might be thinking.   It's a worthwhile exercise only to the extent that a human reader sees the AI analysis, and catches some insights that hadn't occurred to him/her but that make sense when you see them.

See the AI Assessment in SIDEBAR

I've put the AI discussion in a sidebar.  The first AI answer handled the 2025 questions alone.  The second AI answer (clearly marked) was based on the three documents written over 5 years.

https://bqwebpage.blogspot.com/2025/09/moldx-2025-questions-for-barrett-cac.html

Get the 2021, 2023, 2025 PDFs in one zip file here:

https://drive.google.com/file/d/1ptfthX3Uv1yQBLW-jcENnmsJDwwAw9w7/view?usp=sharing


IBEX and Her2 as Killer App. But: The Creaky State of Dig Path Coding and Coverage.

HEADER:

BOSTON, MA, September 4, 2025 – Ibex Medical Analytics (Ibex), a leader in artificial intelligence (AI)-powered cancer diagnostics, today announced it has received European In Vitro Diagnostic Medical Devices Regulation (IVDR) certification for its HER2 breast cancer biomarker scoring solution. 

Ibex’s fully automated “zero-click” AI-enhanced decision support tool for pathologists efficiently increases the accuracy and consistency of HER2 immunohistochemistry (IHC) scoring, including HER2-low cases. Ibex’s breast HER2 solution was developed and validated by Ibex in collaboration with AstraZeneca and Daiichi Sankyo.

###

  • See Linked In heads up from IBEX here.
  • See full IBEX press release here.

One of my most popular posts in 2022, with about a thousand views, asked if Low-Her2 analysis would be a "killer app" for digital pathology.   Here; see 2023 follow up here.  It sounds like the IBEX product will live up to that hope.   (We don't know IBEX's FDA status for Her2 interpretation; IBEX has had breakthrough FDA status in the past, though.)

###

The state of coding and pricing and coverage for digital pathology in the U.S. seems slower.  Up until July 2024, the AMA CPT was creating digital pathology codes in the PLA code system, but that workstream seems to be frozen after July 2024.   (AMA CPT will be discussing AI coding at its September 2025 meeting, including a new proposed category called CMAA or "Tab 94.")

By my tally, there have been 7 PLA codes in the WSI/digital interpretation field (0108U, 0261U, 0376U, 0414U, 0418U, 0512U, 0513U), most of which have been priced at $706.25.  some of those codes include IHC slides, some are based on straight H&E WSI and digital algorithms.

However,  as of 2024 data, utilization at Medicare has been limited.   Of the seven codes, 0108U was paid $27M for 5,500 services, followed by 0261U being paid $615,000 for 870 services, followed by...well, nothing.   As far as coverage rules, I couldn't find any yet for 0108U (e.g. an LCD or other article).

For codes and utlization data see cloud spreadsheet here.


Friday, September 5, 2025

Van Calster (2025) Why Adoption of Clinical Predictors Is So Low

 Annual Reviews of Statistics just released a paper with an interesting title:

The Enemies of Reliable and Useful Clinical Prediction Models: A Review of Statistical and Scientific Challenges.  Here.

Writing,

The current status of applied clinical prediction modeling is poor. Many models are developed with suboptimal methods and are not evaluated, and hence have little impact on clinical care. We review 12 challenges—provocatively labeled enemies—that jeopardize the creation of prediction models that make it to clinical practice to improve treatment decisions and clinical outcomes for individual patients.

##

While the focus is on prediction algorithms, many of the challenges map easily onto the adoption, or disappearance, of diagnostic tests as well (e.g. MAAA predictors).   The need for impact assessment and implementation studies is highlighted, topics close to genomics.


Their summary figure is worth highlighting (from Royen):


##
Van Calster discuss use of AI-ML in Banff transplant classification (p. 10), citing Farris 2023.

Bipartisan Policy Center: Report: Rise of AI in Federal Health Agencies

 On August 10, 2025, the Bipartisan Health Center released a detailed report, with numerous graphics, on "Mapping the Rise of AI in Federal Health Agencies."

It's cited in the briefing memo for the September 3, 2025, House Health hearing on AI in healthcare.

Find the full report here:

https://bipartisanpolicy.org/blog/mapping-the-rise-of-ai-in-federal-health-agencies/


AI Corner

Federal health agencies are rapidly expanding their use of artificial intelligence, moving from pilot projects to core operations in areas like fraud detection, outbreak surveillance, and regulatory review. Under new White House and HHS mandates, agencies such as CDC, CMS, and FDA now track and disclose hundreds of AI use cases, with applications ranging from internal workforce support to high-impact public services. While tools like FDA’s new generative AI assistant Elsa promise major efficiency gains, they also raise questions about transparency, safeguards, and the role of AI in sensitive regulatory decisions.

##

Here’s a TL;DR of the Bipartisan Policy Center report Mapping the Rise of AI in Federal Health Agencies (Aug 10, 2025):

Federal AI Expansion

  • FDA launched Elsa (June 2025), a secure generative AI chatbot for staff productivity.

  • HHS has tracked AI use since 1996, with cases quadrupling between 2022–2024 (270 cases across nine agencies).

  • Growth accelerated after generative AI tools (e.g., ChatGPT) emerged.

Policy & Oversight

  • 2020 executive order required agencies to inventory AI uses.

  • April 2025 OMB memoranda mandated Chief AI Officers, transparency, and standardized evaluation.

  • Agencies must now report AI tools’ purposes, safeguards, and metrics.

Agency Highlights

  • CDC: Uses AI for literature reviews, outbreak prediction, and internal operations (ChatCDC).

  • CMS: Applies AI to fraud detection, claims monitoring, and customer complaints; recent AI-supported fraud takedown blocked $4B in false claims.

  • FDA: Uses AI for data extraction from submissions and document review. Elsa supports writing and summarization but raises concerns about reviewer influence and legal liability. FDA also launched two AI councils: one for regulating AI-enabled products, another for internal oversight.

Key Applications (6 categories):

  1. Education & Workforce (training, chatbots).

  2. Emergency Management (social media monitoring, outbreak tracking).

  3. Government Services (benefits delivery, adverse event monitoring).

  4. Health & Medical (surveillance, real-world data).

  5. Law & Justice (fraud detection, case routing).

  6. Mission-Enabling (internal workflows like drug labeling review).

Takeaway
AI adoption in federal health agencies is accelerating, with a focus on transparency, fraud prevention, regulatory review, and workforce support. Success will depend on governance, safeguards, and clarity around AI’s role in high-stakes regulatory decisions.

 

Subscribe to: Comments (Atom)