Clinical Behavior in Genomics: New Paper on Reflex Tests for Genomic Biomarkers

An important new paper is just released, and open access.  See Pineualt et al., on protocols and barriers regarding reflex testing for gene panels in non small cell lung cancer (NSCLC).   

Find it here:

https://pubmed.ncbi.nlm.nih.gov/42467547/

i've linked via PubMed which gives you, at bottom, a useful list of related papers.

##

AI CORNER

##

What Happens After the Guideline? Achieving Access in Real-World Practice

##

Summary

Pineault and colleagues examine how US health care institutions actually implement comprehensive biomarker testing for non–small cell lung cancer—not simply whether professional guidelines recommend it. Using a national cross-sectional survey of 111 health care professionals involved in NSCLC testing, the study evaluates locally standardized protocols, reflex ordering of broad multigene panels, the role of pathologists, and the regulatory and operational barriers encountered in everyday practice.

The findings are encouraging but reveal a striking implementation gap. Nearly 78% of respondents reported that their institutions had standardized comprehensive biomarker-testing protocols, and 88% of those protocols included reflexing to a multigene panel. At institutions with such protocols, most respondents said that at least 80% of newly diagnosed patients received comprehensive testing. Yet the proportion reporting that results were available before the first oncology visit—when first-line treatment is selected—fell to 47.7%. Thus, adopting a protocol does not automatically ensure that its intended clinical benefit reaches the patient at the critical decision point.

Pathologists were commonly the ordering providers, reflecting their practical position at the center of specimen selection, tumor assessment, tissue stewardship, and testing logistics. More than 90% of respondents believed that multidisciplinary, pathologist-ordered reflex testing improves care, and 93.7% supported CMS recognition of pathologists as ordering physicians, generally within guideline-based or multidisciplinary safeguards.

The study’s greatest contribution may be its treatment of regulation as something institutions must operationalize rather than merely obey. Reimbursement restrictions, prior authorization, payer denials, the Medicare 14-day rule, differing Medicare Administrative Contractor requirements, staffing shortages, and LIS limitations all shaped how—or whether—reflex testing worked. Institutions consequently developed locally adapted arrangements involving pathologist orders, oncology standing orders, multidisciplinary approval, and other hybrid workflows.

This is an important and highly promising approach to health-policy research. By studying real-world institutions, it shows how regulatory policies are translated, negotiated, accommodated, and sometimes worked around inside functioning care systems. The survey is modest, self-reported, and weighted toward pathology professionals, so it cannot establish causality or national prevalence. Nevertheless, it provides precisely the kind of grounded institutional evidence needed to understand why apparently sound policies succeed in some settings but falter in others—and how regulation might be redesigned to support effective care rather than inadvertently obstruct it.

##

About LUNGevity

The study was funded by LUNGevity Foundation, a leading US lung-cancer nonprofit supporting research, patient education, advocacy, and community services. Its precision-medicine initiatives promote timely, comprehensive biomarker testing and work to remove reimbursement, prior-authorization, and regulatory barriers that prevent patients from receiving the right treatment at the right time.  

Very Brief Blog: CMS "Rapid Benefit Categories" for HCPCS Codes

Several years ago, CMS announced a process where it would codify something it was already doing - making benefit category decisions for new HCPCS code applications.   That is, CMS already for years would deny a new HCPCS applicant on the grounds it was "not a benefit category" - "Thanks, CMS."  And if it gave you a code, it usually meant, they believed it did fit some benefit category.

Such decisions are now organized in one place, and regularly updated.  

Hint: you might find the denials more interesting that the acceptances.  Only problem is, most of the denials for "no benefit category" never get a code, either, so they never survive long enogh to make it onto this list.  

For an exception, see "artificial saliva" A9154, p 8, which has no benefit category (although it did get a code!, which probably took a lot of pushing and repeat visits with CMS.)  

Breast milk bags are "contractor discretion" with a manual pump but no benefit category with an electric pump.   (Head-spinning.)

See transmittal here:

https://www.cms.gov/files/document/r13889bp.pdf

CMS Proposes to Pull WSI Tests Off the CLFS—Entering Topsy-Turvy Land?

In both the summer proposed hospital outpatient policies and the proposed physician payment policies, CMS proposed taking software-intensive whole-slide imaging (WSI) tests off the Clinical Laboratory Fee Schedule (CLFS).

Forget the current payment amounts for a moment. CMS proposes that, under Part B, WSI tests would be contractor-priced—eek! In the hospital outpatient setting, where contractor pricing is generally verboten, CMS would temporarily assign the WSI codes to APCs with payment rates roughly similar to what they previously received under the CLFS.

Analysis

For me, the central question is whether these tests are, or are not, CLIA laboratory tests.

If they are clinical laboratory tests of the laboratory type (not the physician pathologist type) then they are paid under the CLFS and governed by CLFS pricing rules, including the PAMA framework. See Social Security Act §1834A. Plain English, no wiggle room.

I do not think CMS can simply say: “Yes, these are clinical laboratory tests, but despite §1834A, we no longer like pricing them on the CLFS.”

But if CMS removes the codes from the CLFS on the theory that they are not CLIA laboratory tests, we enter topsy-turvy land.

###

Medicare Doesn't Define Laboratory Test!

Medicare doesn't have its own definition of (clinical) laboratory test.  PAMA (SSA 1834A) added the term CDLT Clinical Diagnostic Laboratory Test, but no definition.   One assumes therefore the relevant definition is that of CLIA, which is a bulky paragraph that makes any human tissue, fluid, etc, and any analysis method, a CLIA test.  

Code Application (AMA) versus Test Category (Medicare)

One of the craziest examples is that Category III codes for digital pathology will be reviewed by CAP, and Pathology Coding Caucuse, require proof a lab offers the test, require a CLIA license, in a new proposal, even require a letter from the CLIA Lab Medical Director.

Yet, CMS has proposed that digital pathology tests are NOT laboratory tests, NOT CLIA tests, at all.  And if CMS doesn't classify the codes as CLIA tests, then laboratories can't bill them (there are national CMS claims processing edits that labs can only bill lab codes, not, say, MRI of the head).   

Date-of-Service Rules

If they are not clinical laboratory tests, then WSI tests presumably would not be subject to the laboratory date-of-service rules. See next point.

OPPS and Even Inpatient Bundling

If they are no longer subject to the strange, sometimes backward-looking laboratory date-of-service rules....then they may no longer be bundled into inpatient and outpatient hospital payments in the same way CMS takes for granted today.

Codes like 0220U, a digital pathology code, have NOT been payable in the hospital outpatient setting, because they carried status indicator (SI) Q4.  Now, they are proposed to be switched to status indicator O1, and paid $750 in the same setting.

ADLT Status

If they are clinical laboratory tests, they may be eligible for Advanced Diagnostic Laboratory Test status and ADLT pricing.

If they are not laboratory tests, they presumably are not eligible.

PAMA

If they are not clinical laboratory tests, they would not be subject to PAMA pricing and reporting rules.

Enrollment

Today, an entity performing CLIA tests obtains a CLIA certificate and enrolls in Medicare as a clinical laboratory.

But if H&E-based WSI-AI tests are not CLIA tests, can the entity performing them enroll as a clinical laboratory—or not?

As I noted in an earlier blog, CMS initially would not let HeartFlow enroll in Medicare at all. HeartFlow, now a major public company, was eventually permitted to enroll as an independent diagnostic testing facility, or IDTF.

Billing by a CLIA Laboratory

CMS is proposing that certain WSI tests should no longer be CLFS tests. That seems difficult to reconcile unless CMS also regards them as no longer being CLIA laboratory tests.

But my understanding is that a laboratory enrolled in Medicare as a clinical laboratory can bill laboratory services—not E&M visits, or foot surgery, or an MRI of the head.  These are provider-to-code hard wired edits, where the provider must be of the CLIA type or hold a CLIA certificate  to bill the CLIA codes.

If CMS no longer regards certain PLA-coded WSI services as CLIA laboratory services, CMS may be able to remove them from the CLFS. But would CMS also have to remove them from the listing of services that an enrolled CLIA laboratory is permitted to bill?   The lab would still be a CLIA lab but the WSI code it invented and got coded, would get yanked off its billable codes list.  

70/30 Rule - and Other Reference lab rules.

Labs can bill for reference lab tests that they refer out to a specialty lab, if such referrals are not more than 30% of the lab's volume.  (This is to avoid the lab becoming a "paper shell" that doesn't run a lab but only refers tests out.)  However, if digital pathology codes are not CLIA tests, then presumably they wouldn't count toward the 70/30 rule.  ...But, if they were not CLIA tests, then a CLIA lab probably is blocked from billing them.  And if the digital pathology are kept on the CLIA list at CMS so the lab where they are run can bill them, then those tests should be priced by CLFS per statute at 1834A, but CMS wants them off the CLFS and contractor priced...   topsy turvy land, again.

Another aspect, aside from 70/30 rule, when labs acquire a specimen and refer it to a special reference lab (like Mayo or ARUP), there are special claims rules, modifiers, putting the CLIA or NPI of the CLIA performing lab, etc.   None of these would apply if digital pathology is "not a lab test" in the first place.

Billing by an IDTF?

IDTFs generally are prohibited from billing CLIA laboratory services.  That's old news.

But CMS now appears to characterize these WSI services not as clinical laboratory tests, but as general diagnostic tests covered under Social Security Act §1861(s)(3).

If they are general diagnostic tests under §1861(s)(3), and CMS formally treats them that way, then perhaps they could be billed by an IDTF—even though the service involves a glass slide, tissue, and a microscope.

Level of Supervision

CMS maintains a table where it assigns "level of supervision" to all tests which are not CLIA tests.  Pathology tests are exempt from this table, because by definition they are under general supervision.  But CMS is saying that digital pathology tests are no longer CLIA tests (neither of the clin lab type nor the pathologist type), but rather, are general tests under 1861(s)(3), so they will have to be assigned levels of supervision.

###

Head-spinning stuff.

There may be additional consequences that have not yet occurred to me.

 

CLIA, and Computational Pathology on H&E: The Dog that Didn't Bark?

There's a Sherlock Holmes story where a dog barked at every stranger.  Since the murder was committed on a night the dog didn't bark, Holmes concluded the murderer was someone known to the household.

Here, I'm putting together some puzzle pieces from recent CMS (and FDA) decisions.

Background Facts

In July proposed rulemaking, CMS proposed to take  existing whole slide imaging codes, priced on the CMS clinical lab fee schedule and falling under its Clin Lab policies like Date of Service rule, ADLT rule, etc.   And take those codes OFF the clinical lab fee schedule, and price them differently (temporarily on APCs in the OPPS setting, and by "contractor price" elsewhere.)

For me, the natural position was that those WSI tests - including advanced computational results from H&E - were CLIA tests, and simply had to go on the Clin Lab Fee Schedule - nothing to talk about.

For example, see a position paper from Digital Pathology Association on validation of AI "in the clinical lab" - a CLIA lab.  Here, here.  (Link 2, the paper, is open access, but you might need to enter from Link 1).

But CMS has pulled a range of WSI based codes (whether H&E or IHC) from the CLFS in its proposed rules, while skipping quite a few similar codes in the PLA system.  (See Valar comment to CMS, link in this blog.)

Back to the Non Barking Dog

So anyway, here's my point about that non-barking dog.

When FDA approves lab tess (including big sole-lab tests like FMI Foundation One), it classifies them for CLIA.

FDA explains how it classifies IVDs in CLIA levels

 https://www.fda.gov/medical-devices/ivd-regulatory-assistance/clia-categorizations

CLIA explains how this is FDA's job   

https://www.cms.gov/medicare/quality/clinical-laboratory-improvement-amendments/cartegorization

FDA clia database is here

https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfCLIA/search.cfm

If you search that database for Foundation Medicine, you'll find a number of tests. But look for Artera's tests - including their prostate test from 2025 based on H&E algorithm - and you won't find it.

So long before CMS proposed to knock H&E WSI algorithmic tests off the CLFS - which would require they "not" be CLIA tests - FDA was NOT putting them into the national  FDA-CLIA database of FDA tests that were classed by FDA as "CLIA" tests.

It's just an isolated finding but I found it intriguing.

Hiding in plain sight - like the dog that didn't bark.  There's apparently been some hesitation since 2025 whether H&E WSI tests - although for FDA "medical devices," might not also be "CLIA tests" in the CLIA database.

in contrast, for example, at the FDA-CLIA portal search "Dako," you'll get 81 hits.  Search "Abbott" and you'll max-out the system at 500 hits.

CMS Denial Rates for PLA Codes

 CMS has a terrific database where you can see all codes (services) paid for any US laboratory.  Find it here.

CMS also has a database for all CPT codes paid to Part B providers and labs, along with denial rates.  These tables are national, and are not provider specific.  But, if the code is a sole-lab PLA code, it's almost the same as provider specific.  Database here.

I ran a search for PLA codes 037U (FMI CGP Paraffin) and 0239U (FMI CGP LBx).

The table breaks the codes up into multiple lines for obscure reasons (e.g. some will go through the "Palmetto Railroad Retiree Contractor.")   See below for the fragmented lines.

But the data count in the main rows here (with around 13000 cases for 0037U and around 7000 cases for 0239U0) match up with other sourcecs for 2024.

The FMI PLA code denial rate appears to be... about 4% in those main rows.  QED.

Click to enlarge.

click to enlarge

##

##

##

A person could search the data for all codes ending in "U" and get several hundrred rows in Excel.  You could then sort for most popular PLA codes, most denied PLA code, most denied dollars per PLA code, etc.   
 Which I did here as an initial experiment.

AI Guest Author: New 20-page AI White Paper on Medicare vs Commercial Coverage for Genomics

 It's become almost a cliche' in the past decade that for many tests, Medicare coverage will occur ahead of most commercial insurance coverage.    Or, in clinical areas like minimal residual disease (MRD) or comprehensive genomic profiling (CGP; CMS NCD 90.2), Medicare coverage will be broader.

This 15-page white paper was written entirely by Chat GPT, which was given only a few sentences of initial guidance (prompt shown on page 14).  Therefore, it should be taken as an example of the current state of AI planning, research, organization, and writing - rather than a a truth standard or as a reference article.

Find the white paper here.

Capsule Summary

Medicare has repeatedly moved first in advanced oncology genomics: nationally for FDA-linked comprehensive sequencing and locally through MolDX for tumor-informed MRD. Commercial coverage remains fragmented, with several national policies still negative and only positive outliers. Evidence, guidelines, FDA labels, and contracting—not analytic performance alone—will determine the next wave. 

Executive Summary

This white paper finds an ongoing gap between Medicare and commercial insurance in advanced oncology genomics. Almost ten years ago, 2018, CMS NCD 90.2 created national coverage for qualifying FDA-approved or cleared next-generation sequencing companion diagnostics in advanced cancer and preserved contractor discretion for additional tests. That framework was far more permissive than commercial policies of the day.  The pattern recurred in molecular residual disease: MolDX established a test-and-indication pathway enabling broad Signatera coverage --while several major commercial policies continued to classify solid-tumor MRD as unproven, investigational, or not medically necessary.  

The 2026 sample is fragmented rather than uniformly negative. UnitedHealthcare, Aetna, Cigna, Carelon-administered plans, the Federal Employee Program, and Excellus remain adverse; Blue Shield of California is unusually expansive; Centene covers selected Signatera uses; and Arkansas Blue Cross covers a narrow Merkel-cell indication. This dispersion reflects different evidentiary thresholds, especially the gap between "prognostic validity" and actual proof that test-directed management improves outcomes.  

Commercial convergence is likely to proceed no faster than indication by indication. FDA’s 2026 Signatera companion-diagnostic approval in muscle-invasive bladder cancer and guideline changes may accelerate that use first. Laboratories should fund prospective utility trials, pursue regulatory and guideline milestones, constrain testing cadence, and build payer-specific economic evidence. Early movers have a moat, part of which is data, but part of which is the high level of payer resistance. 

Linked In Deep Dive: Daniel G's Writings on Where MRD Is Going

 

Daniel G. and the Rise of the LinkedIn Topic Expert in MRD

https://www.linkedin.com/in/daniel-giner/

This blog written by Chat GPT 5.6.

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

LinkedIn has developed its own class of expert: the experienced industry participant who chooses a relatively narrow field, follows it almost continuously, and publishes enough commentary that readers begin to rely on that person as an informal editor of the topic. These writers are not necessarily producing original clinical research or lengthy investment reports. Their value lies in selecting developments, putting them into a recognizable framework, and returning often enough that a following accumulates.

Daniel G. is a good example in molecular diagnostics and, increasingly, measurable or molecular residual disease—MRD. His profile reports more than 5,400 followers and describes his current work as business development and commercial partnerships in diagnostics, precision medicine, and international expansion.

Big News: The FIND RCT Study in Colorectal Cancer: Integrating ctDNA and Imaging Management (Mo et al.)

SUMMARY:

The randomized phase III FIND trial integrated serial ctDNA methylation testing with protocol-driven CT imaging after colorectal cancer surgery. ctDNA-guided surveillance detected recurrence a median 3.9 months earlier and doubled curative-intent treatment among patients who relapsed (48.1% versus 23.6%). 

• The study moves MRD beyond prognosis toward actionable surveillance, although mature overall-survival, economic, and independent replication data remain essential for adoption.

CITATION:

Mo S, Zhou C, Ma M, et al. Dynamic circulating tumor DNA methylation monitoring guiding postoperative surveillance in nonmetastatic colorectal cancer: a prospective, randomized, phase III FIND trial. J Clin Oncol. Published online July 29, 2026. PMID 42525894.  doi:10.1200/JCO-25-03009.

https://pubmed.ncbi.nlm.nih.gov/42525894/

The article below is written by Chat GPT 5.6. 

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

The Blood Test That Tells the CT Scanner When to Wake Up

The randomized FIND trial moves ctDNA beyond prognosis and into a working surveillance algorithm—doubling the proportion of recurrent colorectal cancers treated with curative intent

Patrick Adams recently highlighted a deceptively simple question about postoperative colorectal cancer surveillance: can ctDNA tell us more precisely whom to image and when, so that recurrent disease is found while it is still manageable? His LinkedIn note is not independent evidence, but it captures the operational insight of an unusually interesting new trial rather well.

The phase III FIND trial, published July 29 in the Journal of Clinical Oncology, did not merely add another biomarker measurement to an observational registry. The investigators built a complete, protocol-driven system in which serial ctDNA results changed the timing of CT imaging, subsequent negative results could turn intensified surveillance back down, and the clinical endpoint was not simply “earlier detection.” It was whether patients whose cancer recurred could receive metastasis-directed treatment with curative intent.

That is a much more consequential test of molecular residual disease.

Linked In Deep Dive: Josh Bowerman's Articles on Genomics Industry Dynamics

Josh Bowerman Maps the Rapidly Changing World of Clinical Genomics

[This blog written by Chat GPT 5.6]

Josh Bowerman has developed a distinctive body of commentary on clinical genomics, particularly oncology diagnostics, liquid biopsy, early cancer detection, comprehensive genomic profiling, and minimal residual disease. His articles are brief, visually polished, and generally published on LinkedIn, but taken together they amount to something more substantial: an evolving map of how the precision-diagnostics industry is reorganizing itself.

Bowerman is not primarily reviewing individual scientific papers or comparing the analytical sensitivity of competing assays. His recurring questions are commercial and strategic:

Who is building what? Which parts of the cancer journey are companies trying to control? Are they developing capabilities internally or acquiring them? And, once the technology works, who can obtain reimbursement and achieve clinical adoption?

Bowerman writes from JBAndrews, an executive-search firm active in diagnostics and precision medicine. His LinkedIn profile, available here, reports approximately 29,000 followers. His position gives him an unusual source of market intelligence: conversations with companies about what expertise they are hiring, what capabilities they lack, and where they expect growth. In one discussion, he says explicitly that the firm’s view of deal activity often emerges from recruitment conversations.

These are therefore not neutral technology assessments or systematic evidence reviews. They are better understood as market cartography: concise narratives and graphics showing how a fragmented field may be consolidating.

LinkedIn Deep Dive: Jack Kohler's Work on Interpreting Clinical Trials for Commercial

 

Reading the Clinical Paper Commercially:
Jack Kohler’s Methods to Bridge from Evidence to Strategy

The essay below is written by Chat GPT 5.6 after reading Kohler's web articles.  He's focused on biopharma, where one main pivotal trial usually is hugely influential in FDA decisions as well as coverage.  In diagnostics, VERY often, at least tests in novel areas, a whole sequence of papers are required for coverage.   E.g. see the lengthy clinical trial page for Naveris NavDx for MRD in H&N cancer.  Here, here.

# # # # #

Clinical papers are written to report research. Pharmaceutical companies must ask them to do something more: clarify which patients matter most, whether a product is meaningfully differentiated, what claims the evidence can support, and whether the result can become a credible positioning and value story.

Jack Kohler has built a consulting practice around that translation. His background spans pharmaceutical sales, brand management, and international marketing, and he describes his work as helping pharma and biotech teams turn clinical evidence into “decision-ready strategy.” (LinkedIn)

His recent infographic, “How to Read a Clinical Paper Commercially,” organizes the task into nine steps:

Major News: FDA Posts Its 28pp Review of ArteraAI - Breast Cancer Computational Pathology

In May 2026, ArteraAI got FDA clearance for its computational pathology test for breast cancer.  FDA can take "weeks to months" to post its 20- to 30-page review packet - and the breast cancer review is now public.

Artera AI Breast Cancer (NEW)

See the K254114 product home page here;

https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfPMN/pmn.cfm?ID=K254115

See the FDA's 28-page 510(k) review here:

https://www.accessdata.fda.gov/cdrh_docs/reviews/K254115.pdf

FDA notes that the application includes a Predetermined Change Control Plan PCCP for updates.

See the four-page FDA letter here:

https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfPMN/pmn.cfm?ID=K254115

See the 864.3755 regulatory classification here, product classification SHW:

https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpcd/classification.cfm?id=5520

Pathology software algorithm device analyzing digital images for breast cancer prognosis

AI Guest Author: AI Researches, Writes, White Paper on the "87798" Story at Medicare

A once-obscure CPT code, 87798, PCR for "other organism," exploded in usage to become the highest-paid molecular code in 2014.  During 2016, OIG reports have appeared and Medicare MACs rapidly began to slash payments for 87798.  See a representative Linked In article, here.

As an experiment in AI writing - I briefly asked Chat GPT 5.6 to research 87798 at Linked In at elsewhere, get 20-30 references, and then write a white paper.   That is, my instructions were minimal, and the project was AI-based research and writing.

Here is the 16-page PDF white paper:

https://drive.google.com/file/d/1qGAjKawZyn3n7OCL96IoNVgdyfUDKHOz/view?usp=sharing

Very Brief Blog: New Book Coming: Pixels and Precision Pathology

 It's not out yet, but I'm seeing publicity for a new book on computational pathology, "Precision Pathology: From Pixels to Predictions."  Springer, here. 

The Amazon page says it ships November 4, 2026, here.

See LinkedIn for one of the coauthors, Dr. Gu of Pittsburgh, here.

If that's of interest, you may also like Rajni Toppo's essay, Human vs. Machine Judgment in Pathology: Rethinking the “Human Baseline” - here.

And you might like a new article in NEJM AI, "Tranlating AI in Pathology into Clinical Impact" (Su et al., 2026).

Back to the "Precision Pathology" book, here is a quick AI summary:

Precision Pathology: Pixels to Predictions captures pathology’s transformation from microscope to computational medicine. Grounded in real-world workflows, it explores digital pathology, AI, foundation and agentic models, and multi-omics. 

Its concept of “pixelomics” traces the journey from pixels to patterns, diagnosis, prediction, and molecular insight—offering pathologists, trainees, and researchers a practical guide to the emerging future of precision pathology.

##

##

Also in today's news, a long headline article on AI from NYT (July 29):

https://www.nytimes.com/interactive/2026/07/29/technology/ai-chips-data-center-boom.html

The Times describes an unprecedented global build-out of AI computing infrastructure. About 20 million advanced chips support AI today, potentially rising tenfold by 2028, as U.S. tech giants spend hundreds of billions on data centers. Advocates expect accelerating breakthroughs, agents, and automation; critics warn of bubbles, job disruption, soaring electricity demand, environmental impacts, widening inequality, and intensifying U.S.-China geopolitical rivalry.

 

  

AI Corner: AI Counts New Diagnostic LCDs, Timelines

I was interested in how many LCDs the different MACs produce for lab diagnostics (MolDx, Novitas, Wellpoint Federal [former NGS MAC].)

To my surprise, Chat GPT was able to follow instructions, search the CMS database for different parameters, and produce clean tables as output.  It could also volunteer insights along the way.

You can see my long dialog with Chat GPT in a PDF, and also get Excel tables for MolDx, Novitas, Wellpoint, in this cloud zip file.

https://drive.google.com/file/d/1ASG_aO-28uEDD8qiujrb4OzUTqFitxlx/view?usp=sharing

MolDx

For a 3-year period, it pulled 7 LCDs, although only one or two (L39583, Squamous Risk Stratification) were for wholly new tests.  Some LCDs brought older coverage under one new foundational policy.  A couple were for well-known guideline hereditary genes e.g. L39953, genes for thoracic aorta designs.   These are important LCDs, but if YOUR test is some "wholly novel idea," you'd be asking, how many such LCDs they put out, not which genes are covered for aortic aneurysms.

Of course, MolDx specializes in broad "foundational" LCDs, so new tests from new labs are added on a rolling basis via Z code activation.

Regarding timelines, request received to LCD draft released, shows ~800 days was typical (range, 724-1494 days).

Novitas

At Novitas, Chat GPT followed the same template and found just one new diagnostics LCD, L39365 which non covers 9 out of 9 proprietary LDT oncology tests. 

There was no request-to-final time, as it was internally generated. However, the record does show it dragged through comment and revisions for about 30 months, a kind of timeline check.

Wellpoint

Chat GPT tallied 3 LCDs at Wellpoint, l39611 for urine drug testing, L39995 for PGx (very similar to MolDx LCD for PGx0, and l39726 for "KidneyIntelX."   Only the third, KidneyIntelX, would be of interest as a "wholly novel test" LCD.

As to timeline, none listed a formal request-letter of record.  However, clever AI reports that Renalytix publicly stated it had submitted an LCD request in October 2022, and the LCD was finaled in June 2024,  so we can infer about 20 months from request to final LCD.  

There was a CAC along the way, in August 2023. Code 0407U for KidneyIntelX had about 250 CMS payments in 2024.   RENX 2018-2025 raised about $220M for a current market cap of $15M.

Take Home Lesson for Innovators

In the three years studied, Chat GPT (working on its own) reports 9 LDT non coverage decisions at Novitas (1 LCD); 1 novel test LCD at Wellpoint, and several novel LCDs (like one on squamous cancer MAAA) at MolDx.   As noted, counting at MolDx was more complex; guideline-matching hereditary gene coverage probably is not germane for someone with a new-out-of-the-box creation.

click to enlarge

click to enlarge

Nerd Note: How CMS Handles Initial OPPS Pricing of the Newest Cat III Codes

In the previous blog, I noted that CMS provides proposed OPPS/APC prices of Cat III codes that won't be effective for six months...and that don't even have public final code numbers.  (The same blog shows a comment letter from Valar that includes both the placeholder codes and real codes side-by-side).

This is a Chat GPT explanation of how that particular bit of CMS magic works.

###

CMS Prices New CPT Category III Codes
Before Their Final Code Numbers Are Public

Yes it does! And CMS explicitly discusses how it handles new CPT Category I and Category III codes that will take effect January 1, 2027 even though their final CPT numbers have not yet been released publicly.

The key discussion is in the CY 2027 OPPS Proposed Rule at 91 Fed. Reg. 41787–41788, Section III.A.4.b, “New CPT Codes Proposed Rule Comment Solicitation.”

CMS explains that it receives the upcoming January CPT codes from the AMA early enough to use them in the OPPS proposed rule:

“For the CY 2027 OPPS update, we received the CPT codes that will be effective January 1, 2027, from the AMA in time [by June 1?] to be included in this proposed rule [with proposed prices].”

However, because the permanent CPT numbers are not yet available for use in the proposed rule, CMS uses five-character AMA/CMS placeholder codes. Thus codes such as X568T, X569T, X614T, X623T, and X624T appear in the proposed OPPS files even though those are not the eventual CPT numbers.

CMS divides the information between two addenda.

Addendum B [dollars] contains the placeholder code, a short descriptor, the proposed OPPS status indicator, APC assignment, and therefore the proposed payment.

Addendum O [words] contains the placeholder code together with the full long CPT descriptor. CMS specifically explains:

“Therefore, we are including the 5-digit placeholder codes and the long descriptors for the new and revised CY 2027 CPT codes in Addendum O, specifically under the column labeled ‘CY 2027 OPPS/ASC Proposed Rule 5-Digit AMA/CMS Placeholder Code.’”

Nerd note:  Appendix O has nearly 200 codes of different types (Cat III, also CPT Cat I, G-code, etc) that were in some state of partial definition as the rule went to press in June. 

CMS then states that the final HCPCS/CPT code numbers will appear in the CY 2027 OPPS/ASC final rule. Thus CMS is able to propose an APC assignment—and an actual dollar payment—for a new Category III service before the public-facing permanent Category III number has appeared. 91 Fed. Reg. 41787–41788.

CMS's OPPS rule and addenda are available here:

https://www.cms.gov/medicare/payment/prospective-payment-systems/hospital-outpatient/regulations-notices

The Regulations.gov docket for the CY 2027 OPPS proposed rule is:

https://www.regulations.gov/docket/CMS-2026-2344

There is also a parallel discussion for the ASC payment system at 91 Fed. Reg. 41935–41936. There CMS again explains that it has received the January 2027 CPT codes from AMA, uses five-character placeholder codes in the proposed rule, places their complete long descriptors in Addendum O, and will substitute the final CPT numbers in the final rule.

The Valar Codes Provide a Good Example

The Valar comment supplies the subsequent permanent numbers for several of these placeholders:

X568T → 1063T — Vesta Bladder Risk Stratify
X569T → 1064T — Vesta Bladder BCGPredict
X614T → 1097T — Vitara Pancreas ChemoPredict
X623T → 1106T — H&E AI analysis, breast cancer
X624T → 1107T — H&E AI analysis, prostate cancer

Valar also reproduces the full descriptors for its three own codes.

So the somewhat counterintuitive sequence is:

AMA creates and communicates the new CPT code to CMS → CMS publishes it under an Xxxxx placeholder and proposes an APC/payment → AMA's permanent CPT number becomes available → CMS substitutes that final number in the OPPS final rule.

The particularly interesting point for the SaMS discussion is that CMS isn't merely acknowledging these unreleased codes are in play. CMS staff are already making substantive payment-policy decisions about them—including whether an algorithm gets $350.50 or $750.50—while identifying the service publicly only through its temporary AMA/CMS placeholder code.