two path book reviews audio script

 

Two introductions to digital pathology

Luis Cano’s Getting Started with Digital Pathology and Aleksandra Zuraw’s Digital Pathology 101, second edition, offer complementary introductions to a field whose scientific possibilities have advanced faster than many professionals’ ability to evaluate them. Cano provides a concise entry into computational tissue analysis and the professional skills needed to begin. Zuraw offers a substantially broader account of the technology, organizational work, and validation required to make digital pathology function in practice. Readers entering the field can profit from both; readers responsible for implementation will find Zuraw the more substantial foundation.

The distinction matters because “digital pathology” encompasses several different activities: viewing scanned slides, managing diagnostic workflows, quantifying tissue features, and deriving clinically meaningful predictions from images. Both authors recognize these differences. Their books are best understood as professional orientation guides, rather than comprehensive textbooks or substitutes for technical and regulatory source documents.

This review evaluates the supplied 48-page Cano PDF and 179-page Zuraw PDF. Page references below refer to PDF page positions, including covers, rather than printed pagination.

The authors and their perspectives

Luis Cano Ayestas is a medical pathologist with a PhD in oncology and the author of Beyond the Slide, a publication addressing digital pathology, AI, and diagnostic systems. His public LinkedIn listing identifies a Paris-based independent professional and lists Université de Rennes I. His own website describes an interest in connecting histologic interpretation with computational methods and healthcare strategy. In the book, he recounts training in Peru, doctoral study in France, and gradually acquiring skills in R, QuPath, and Python. That trajectory informs his central educational message: clinical expertise provides a useful starting point for computational work, even before a physician develops advanced technical skills. [1–3]

Aleksandra Zuraw, DVM, PhD, DACVP, is a board-certified veterinary pathologist, founder of Digital Pathology Place, and host of the Digital Pathology Podcast. Her professional background includes tissue image analysis, pharmaceutical biomarker work, and toxicologic pathology. Her public LinkedIn listing and author website support that combination of pathology, education, and technology. The book draws extensively on her work with image analysis scientists and on interviews with clinicians, developers, and industry leaders. Her veterinary background gives the review of digital pathology a useful breadth, particularly where human clinical pathology intersects with preclinical drug development. [4–5]

Where the scope differs

Dimension	Cano	Zuraw
Organizing purpose	Help a newcomer understand the opportunities and begin participating	Explain the ecosystem and prepare readers to evaluate and implement it
Technology coverage	Brief WSI introduction; emphasis on analysis software and computational tasks	More extensive treatment of scanners, image management, storage, interoperability, and displays
Image analysis	Concentrated discussion of quantification, biomarker scoring, segmentation, and spatial relationships	More extended teaching of computer vision, machine learning, transfer learning, and foundation models
Practical implementation	Initial skills, resources, mentoring, and small projects	Workflow redesign, validation, software deployment, monitoring, and organizational change
Drug development	Translational research and clinical trial applications	Clinical development plus a dedicated treatment of toxicologic pathology and GLP workflows
Learning style	Concise explanations, takeaway lists, software examples, and personal advice	Conversational explanations, diagrams, repeated case examples, and links to interviews and courses

These differences are not simply consequences of length. Cano spends a large proportion of his short book explaining what can be extracted from tissue images. Zuraw devotes considerably more space to the conditions under which extraction, interpretation, and reporting become dependable institutional services.

Cano’s strongest contribution

Cano is particularly effective at connecting computational terminology with biologic questions. His discussion of cell quantification progresses from counting cells to identifying clinically relevant populations and interpreting their proportions. His treatment of spatial analysis then asks where those cells lie in relation to tumor, stroma, vessels, and one another. Examples involving immune infiltration and tumor microenvironments explain why tissue architecture contains information that a simple positive-cell percentage cannot capture (PDF pp. 20–28).

This is a useful introduction for a reader who understands histopathology but has not yet considered how to formulate an image analysis problem. Semantic and instance segmentation become intelligible through their consequences: identifying a tissue compartment requires a different representation from separating and counting individual nuclei. The discussion of nearest-neighbor distances, clustering, and related spatial measures broadens the reader’s conception of what digital pathology can measure, although it does not teach those methods in sufficient detail to implement them independently.

The software survey also provides an accessible starting point. QuPath, HALO, and Aiforia represent different approaches to access, support, customization, and model development. Cano’s strongest practical recommendation is to begin with available images and a manageable analytic task. His references to public datasets and learning communities make that recommendation actionable without requiring a laboratory to purchase a scanner first (PDF pp. 17–19, 33–38).

His personal account is relevant to the educational purpose. It addresses a recognizable barrier: pathologists may assume they must master programming and neural-network mathematics before they can contribute. Cano instead explains how a clinician can define meaningful questions, annotate responsibly, and collaborate with technical colleagues. For residents and early-career investigators, this may be the book’s most valuable contribution.

The limitations follow from the same approach. The software descriptions provide orientation rather than a rigorous purchasing comparison. The guide is also more encouraging than discriminating in some descriptions of clinical benefit. Statements about improved accuracy, molecular prediction, or reproducibility need to be read in relation to the particular task and validation setting. Success on a research dataset does not establish that an algorithm can replace an indicated molecular assay or improve treatment decisions.

There are also editorial weaknesses in the referencing. The opening definition is attributed to USCAP, while reference 1 identifies the Royal College of Pathologists. A discussion invoking Borowsky’s diagnostic noninferiority work points to a reference listing Barisoni’s nephropathology review. These mismatches do not invalidate the introductory account, but they make verification less straightforward than a professional reader should expect.

Zuraw’s strongest contribution

Zuraw’s principal strength is explaining why a working digital pathology service requires much more than image acquisition or a high-performing algorithm. Her treatment connects scanner performance and image quality with storage, file formats, image management, laboratory information systems, and the pathologist’s workstation. The reader gains a clearer understanding of how a failure in one component can affect the entire diagnostic process (PDF pp. 36–49, 61–73).

The image analysis teaching is also stronger than the title “101” might suggest. Comparisons among object detection, semantic segmentation, and instance segmentation tie technical choices to the information required from an image. Zuraw repeatedly asks whether a simpler method can accomplish the task adequately, and she connects computational complexity with processing time and infrastructure demands. This helps readers evaluate a proposed solution on its intended function rather than the sophistication of its architecture (PDF pp. 97–119).

The most valuable section concerns deployment. Drawing on Andrew Janowczyk’s account of an H. pylori detection tool, Zuraw describes a project in which algorithm development took approximately two weeks and clinical deployment took three years. The lesson is developed through specific requirements: stable software dependencies, documentation, error handling, integration, defined responsibilities, and continued monitoring. Her explanation of a locked model operating within a changing laboratory environment makes performance drift concrete (PDF pp. 142–154).

The dedicated chapter on toxicologic pathology is the book’s clearest distinctive contribution. It explains why cohort-based evaluation across animals and dose groups differs from diagnosing an individual patient, then connects digital workflows with peer review, electronic records, auditability, and Good Laboratory Practice. Readers whose experience is confined to clinical oncology may find this chapter especially useful because it exposes an important application of digital pathology outside routine patient diagnosis (PDF pp. 160–172).

Zuraw also gives more attention to bias, human interaction with AI, generative systems, and data governance. She recognizes that pathologists can reinforce an algorithm’s errors rather than automatically correct them. Her discussion therefore provides a useful counterweight to simpler descriptions of AI as an invariably beneficial second reader.

The tradeoff is a longer and sometimes repetitive book. Cases and motivational themes recur, and references to the author’s courses, podcast, and subscriptions periodically interrupt the exposition. Interview material supplies valuable operational detail, but individual experiences cannot establish how typical an outcome or implementation timeline will be. A graduate-level reader should distinguish those accounts from independent comparative evidence.

Regulatory statements also require careful checking. For example, an early passage states broadly that a remote home office must meet CLIA certification requirements; later passages describe continued flexibility for digital review. CMS’s post-emergency guidance permits qualifying remote digital review under the primary laboratory’s certificate, subject to specified conditions. The book would benefit from a more consistent presentation of that distinction (PDF pp. 24, 28–29). [6]

Important topics that remain insufficiently developed

The most consequential shared limitation is the lack of a systematic framework for judging clinical utility. Both books discuss validation, and Zuraw covers it substantially more thoroughly. Neither consistently separates technical performance, diagnostic accuracy, changes in management, and improved patient outcomes. A prognostic association does not by itself demonstrate prediction of treatment benefit. Likewise, detecting additional abnormalities does not establish that acting on them improves care.

This distinction is particularly relevant to Cano’s statement that the McGenity systematic review confirmed clinical utility. That review evaluated diagnostic test accuracy and emphasized limitations in the underlying evidence. It should not be treated as a general demonstration that AI-guided care improves outcomes. A stronger evidence chapter would teach readers how to assess external validation, patient-level data separation, calibration, treatment interactions, and prospective workflow or outcome studies. [7]

Reimbursement and market access are another substantial gap. Cano mentions Category III digitization codes as emerging reimbursement opportunities, while Zuraw more cautiously emphasizes that digital conversion often creates operational value without separate revenue. Neither supplies a sufficient framework for distinguishing a code’s existence from coverage, payment methodology, or actual collection. CAP describes the relevant digitization codes as reporting mechanisms without assigned RVUs or established payment. A business case should therefore identify the payer and payment setting before assigning revenue to them. [8]

Both books also stop short of a complete procurement and financial assessment. Zuraw discusses total ownership costs and ongoing expenses; Cano points readers toward ROI resources. Neither develops a worked model incorporating scanning labor, rescans, validation time, storage growth, downtime, and achievable changes in case throughput. They give readers useful questions, but not enough information to approve a major investment.

Data governance receives attention, especially in Zuraw, but neither provides detailed guidance on ownership and licensing of images, annotations, derived models, and secondary commercial uses. Similarly, interoperability is discussed without sufficient detail to design and acceptance-test an enterprise integration. Procurement contracts, cybersecurity recovery exercises, vendor exit arrangements, and the allocation of responsibility after an AI error would all require additional sources and specialist advice.

Finally, neither is a hands-on methods manual. Cano encourages experimentation and Zuraw explains concepts clearly, but readers seeking a reproducible analysis pipeline will need software documentation, exercises, and a defined validation protocol. Their common emphasis on histologic whole-slide imaging also leaves cytology and other technically demanding specimen types less fully treated.

Recommendations to readers

Readers seeking a short introduction before exploring quantitative histology should start with Cano. His guide is especially suitable for pathology residents, clinicians entering translational research, and investigators who need to turn a biologic question into a tractable image analysis task. Its value is greatest when reading leads immediately to a small exercise with real images.

Readers evaluating a service, selecting technology, or organizing implementation should give priority to Zuraw. Her broader treatment is more useful to laboratory directors, implementation teams, pathology informaticists, pharmaceutical scientists, and professionals moving between clinical and preclinical settings. Chapters on AI deployment and toxicologic pathology deserve attention even from readers already familiar with basic WSI concepts.

Both books are worth reading for professionals who must connect scientific opportunity with practical adoption: a pathologist serving as the clinical lead on a computational project, a translational scientist moving an image-derived biomarker toward clinical use, or a product manager who needs to understand both tissue analysis and laboratory operations. Cano provides a quick introduction to the questions digital tissue analysis can answer; Zuraw helps the reader recognize the work required to answer them reliably in practice.

For those readers, a sensible sequence is Cano first, followed by Zuraw, with particular attention to validation, deployment, and the regulatory setting relevant to the intended use. A concrete implementation or investment decision should then be supported by current primary guidance, independent evidence for the specific application, and a local financial analysis. The books provide a useful foundation for that work and help readers recognize where further expertise is needed.

Sources and author links

Primary texts: Luis Cano, Getting Started with Digital Pathology, supplied 48-page PDF; Aleksandra Zuraw, Digital Pathology 101: All You Need to Know to Start and Continue Your Digital Pathology Journey, second edition, copyright 2026, supplied 179-page PDF. The supplied Word notes provided discovery links. Biographical context was checked against public LinkedIn search listings and accessible author websites; full LinkedIn profile pages were not retrievable.

1. Luis Cano, author description and book access: https://payhip.com/b/wldfe

2. Luis Cano, Beyond the Slide author page: https://beyondtheslide.substack.com/about

3. Luis Cano Ayestas, LinkedIn: https://www.linkedin.com/in/luiscanoayestas/

4. Aleksandra Zuraw, author biography: https://digitalpathologyplace.com/about/

5. Aleksandra Zuraw, LinkedIn: https://www.linkedin.com/in/aleksandra-zuraw-dvm-phd-dacvp/

6. CMS, QSO-23-15-CLIA, May 11, 2023, pp. 2–3: https://www.cms.gov/files/document/qso-23-15-clia-original-release-2023-05-11.pdf

7. McGenity C et al. Artificial intelligence in digital pathology: a systematic review and meta-analysis of diagnostic test accuracy. npj Digital Medicine. 2024;7:114. https://doi.org/10.1038/s41746-024-01106-8

8. College of American Pathologists, How To Use The New Digital Pathology CPT Codes: https://www.cap.org/article/how-to-use-the-new-digital-pathology-cpt-codes-cap-answers-questions/

 

 

EXTRA

 

I asked Chat GPT if they discuss my thesis for dramatic digital pathology growth – 3 points.

 

 

1.      The supply chain must not end with a DICOM presented to a doctor.  We must present a full report ,with draft micro and draft diagnosis, draft grading (e.g. Gleason), for validation and sign out. 

2.      This software MUST BE FDA APPROVED.  No more horsing around with RUO shareware that must be tediously validated at each location.  Sell sAMD to 3000 hospitals not shareware for 4.

3.      We had to dramatically reform and advance input or interaction devices – a new generation that we can’t even clearly foresee.  But we can do it. 

##

Yes—especially Zuraw. But your thesis goes further than either book, and the leap in input and control devices is largely missing from both. It would make a useful additional lens for the review.

Your argument sets a concrete endpoint for digital pathology: the system should help produce a completed diagnostic work product, ready for pathologist review and confirmation, using commercially supported clinical software and a substantially better interface.

Your priority	Cano	Zuraw
Draft the microscopic description and diagnosis for pathologist confirmation	Compatible with his vision of human–AI collaboration, but not developed into a reporting workflow.	Explicitly discusses report drafting, image-and-text integration, and automated reporting. However, she treats diagnostic generation cautiously and emphasizes supporting applications.
FDA-cleared commercial products that ordinary laboratories can purchase and deploy	Supports regulation and validation, but his practical starting point emphasizes QuPath, experimentation, and acquiring analysis skills.	Much closer to your position: emphasizes production-grade software, commercial development, maintenance, integration, and reliable failure handling. She does not, however, adopt an exclusively FDA-cleared pathway.
A generational advance in how pathologists control and navigate virtual slides	Essentially undeveloped.	Discusses ergonomics, usability, and workflow integration, but does not develop a new model of physical interaction with digital slides.

The reporting connection is quite direct in Zuraw. On PDF pp. 123–127, she discusses drafting synoptic reports from dictated findings, generating structured reports from image inputs, and combining image findings with clinical information. Those are relevant passages.

There is nevertheless an important difference between turning a pathologist’s dictation into a report and examining the tissue and preparing a proposed microscopic description and diagnosis. Your thesis centers on the latter. Zuraw touches that possibility, but does not organize the book around delivering it.

Her discussion of commercial software is the closest match to your frustration. On PDF p. 144, she distinguishes research code that works in its developer’s hands from a clinical tool that other people can use, maintain, and trust. On p. 157, she explicitly identifies commercialization and engineering as necessary to produce dependable products. That is strongly compatible with your demand that laboratories should be able to buy a finished clinical product.

But it would overstate her position to say she demands FDA clearance for every clinical application. She also discusses internally developed laboratory algorithms and substantial local validation. Your position is a stronger product and commercialization requirement. I would also distinguish your objection to laboratories having to finish and substantiate a research product from any suggestion that purchasing cleared software eliminates local implementation checks.

Your interface argument adds something neither book adequately supplies. Zuraw’s treatment of ergonomics on PDF p. 68 concerns monitors, seating, lighting, and software usability. That addresses today’s workstation. Your question is more ambitious: what interaction system would let a pathologist examine, compare, interrogate, and confirm tissue findings with dramatically less effort? Neither author provides that vision.

You do not need to specify the winning device to identify the unmet requirement. Possible directions—my extrapolations, not their proposals—include coordinated voice and physical controls, navigation that anticipates the next relevant field, synchronized movement across stains, and direct movement between a proposed report statement and its supporting tissue region.

My assessment: Zuraw supplies substantial support for your first two priorities; neither book adequately develops your third. Your contribution would be to make all three explicit criteria for success, rather than treating digitization, interoperability, or an accurate standalone algorithm as sufficient endpoints.

 

Original SAMS post Nov 4. Replaced with de novo Claude version Nov 9.

 

Wednesday, November 4, 2026

CMS PULLS ALGORITHMIC LAB SERVICES OFF THE CLFS POLICY SYSTEM: Final Rules CY2027

Find the inspection copy OPPS rule here:

https://www.federalregister.gov/public-inspection/2026-22739/medicare-program-hospital-outpatient-prospective-payment-and-ambulatory-surgical-center-payment

Find the inspection copy PFS rule here:

https://www.federalregister.gov/public-inspection/2026-22743/medicare-and-medicaid-programs-calendar-year-2027-payment-policies-under-the-physician-fee-schedule

###

AI CORNER

###

EXECUTIVE SUMMARY

• CMS has finalized a major change in Medicare payment policy for algorithm-only laboratory tests, effective January 2027. 
• CPT codes previously paid under the Clinical Laboratory Fee Schedule (CLFS) (n=14) will instead be treated as other diagnostic tests.  That means, they'll be paid under rules for the Hospital Outpatient Prospective Payment System (OPPS) or the Physician Fee Schedule (PFS). 
• 
  
• Initial payment rates will approximate existing CLFS rates, but conventional Medicare coinsurance will apply.  
• Laboratory-specific rules, including the 14-day date-of-service rule, will no longer apply. 
• CMS corrected several code classifications following public comments but retained its fundamental policy. 
• Important questions remain about future pricing, packaging, and regulation.

###

CMS Final Rules for 2027: Software-Based Laboratory Tests Move Off the CLFS

   

1. OPPS Final Rule: SaMS Analyses Performed on Laboratory Tests

See CMS-1850-FC, Section X.B.2, esp. Table 85

CMS has largely finalized its proposed policy, with important corrections to the affected CPT codes.

CMS maintains that a stand-alone algorithmic analysis of previously generated laboratory data is not a clinical diagnostic laboratory test (CDLT), even if it has traditionally been coded and paid as one.  That was then.  This is now.

Instead, CMS considers them as other diagnostic test under Social Security Act §1861(s)(3).

This represents a substantial break from the historical CLFS treatment of these services, including the established AMA CPT coding process and CMS crosswalk/gapfill procedures. CMS rejected laboratory-industry arguments that laboratory-derived data interpretation remains intrinsically part of the laboratory service.

Payment and the 14 affected codes

CMS finalized 14 codes, compared with 10 in the original proposal. For CY 2027, they receive new technology APC assignments with status indicator O1, allowing separate payment without multiple-procedure discounting.

The final APC assignments are:

14 CPT codes	OPPS New technology APC	Standard payment
0510U	1506	$450.50
0512U, 0513U, 0414U, 0418U, 0220U, 0376U, 1106T, 1107T, 1097T, 1063T, 1064T	1509	$750.50
81417	1505	$350.50
81427	1522	$2,250.50

Source: OPPS final rule, Table 85, pp. 21–23 of the supplied laboratory excerpt. The amounts are standard APC payment rates, before applicable OPPS adjustments.

CMS bases the assignments principally on CY 2026 CLFS payment rates. Because new technology APCs have fixed payment bands, the new payments approximate ($250, $350, $450, etc).

Corrections to the July proposal

CMS made several important adjustments following public comment that highlighted errors and problems.

• Removed 0511U and 81416: CMS accepted that these services include wet-laboratory work and should remain on the CLFS. Deleted code 0208U was also removed.

• Retained 0414U and 0220U: CMS rejected objections that these services may entail wet-laboratory components, concluding that their descriptors primarily identify algorithmic analyses.

• Added 81417 and 81427: These codes for re-evaluation of previously obtained exome and genome sequences had been omitted from the proposal.

• Raised 1063T and 1064T: CMS changed the bladder cancer algorithm codes from proposed APC 1542, paying $350.50, to APC 1509, paying $750.50, based on comparability to other services.

The treatment of codes involving staining or slide digitization when performed remains an interesting definitional issue. CMS insists its policy concerns stand-alone computational services, while retaining several codes whose descriptors contemplate preparatory laboratory activity.

 

Important consequences beyond price

Beneficiary cost-sharing. These services will enter ordinary OPPS payment rules, including 20% beneficiary coinsurance, rather than remaining under the generally no-copayment CLFS.

The 14-day rule. CMS explicitly confirms that SaMS analyses performed on laboratory tests are no longer governed by the laboratory date-of-service regulation at 42 CFR 414.510. This affects a longstanding issue in separately billing molecular and genomic testing following hospital encounters.  For example, it could be payable the day after inpatient discharge, when a DNA test would still be bundled 14 days. 

An inpatient biopsy on July 1 has DNA extraction on July 3, sequencing on July 5, one type of report signed out on July 7.  The data is used for an algorithm and reported on July 9.  The DOS of the    DNA report is July 1 (date of biopsy) but the DOS of the algorithm test is July 9 (date of performance, no special rule.)  The July report has DOS July 1 and is bundled to DRG, the July 9 AI test is not bundled since it is "not a lab test."

CLIA and FDA. CMS emphasizes that the payment change does not itself revise CLIA requirements. It also distinguishes these laboratory-derived SaMS from the FDA-regulated SaMS devices addressed in the preceding OPPS section: FDA marketing authorization is not currently required for the laboratory-derived category.  

Packaging. For 2027, status indicator for SAMS, O1 provides separate payment, and new technology APC services are protected from comprehensive APC packaging. That is reassuring in the immediate term. Longer-term treatment is less certain, especially given the OPPS tradition of packaging most ordinary laboratory services, apart from molecular pathology and certain other exceptions. See: "APPENDIX - Packaging."

Pricing new tests. CMS finalizes assignment of future new laboratory SaMS codes to new technology APCs but does not establish a comparably detailed initial valuation process. With these servicesnow being outside the CLFS, the usual summer crosswalk/gapfill process is no longer the entry point for establishing their prices. CMS acknowledges the need for further development of its APC assignment methods.

Test design.  There may be some tests that are advantaged by having no included wet lab, other tests where one arbitrarily wants to include some wet lab component to "force" the CLFS/ADLT pathway.

Coding.  Remember for a couple years CPT has taken the position that PLA codes MUST  contain new wet lab biomarkers, so CMS algorithm codes won't generally be under PLA. That is, CMS rule for new wetlab = PLA  rule for new wetlab.   In 2026, CPT has made several Cat III codes for H&E SW LDT.  (And watch in 2027 for CMAA codes, TBD.)

The Words of the Code.  Note that CMS makes its decisions based on "words of the code," even when reference to the actual named test would be quick and clarifying.

Proprietary Codes.  PLA codes can only be open use if FDA approved.  Cat III codes always can be open use (if too specific...), potentially friendlier to RUO software deployed at multiple locations and "codifiable" as long as 1 lab gets a Cat III.  This sounds like inside baseball but could be pretty important in some LDT scenarios.

2. PFS Final Rule: SaMS Laboratory Analyses

See CMS-1848-F, Section (61), esp. Table A-D8

The most significant PFS change from the proposal is CMS's decision to reject contractor pricing.

The proposed rule would have placed the identified algorithmic services under Medicare Administrative Contractor pricing, potentially generating geographic variability and uncertainty.

Commenters objected, pointing out that the underlying services have no obvious jurisdiction-specific cost differences and that the proposal would unnecessarily disrupt nationally established prices.

CMS agreed. The final rule establishes national PFS pricing for the same 14 laboratory SaMS codes, with procedure status indicator A (active).

The conversion from OPPS dollars into RVUs

CMS uses the OPPS new technology APC payment as the starting point. It then calculates the practice expense RVU component by dividing the OPPS payment by the CY 2027 PFS conversion factor.

 

In broad terms, a service with an OPPS payment of $750.50 is translated into RVUs using the approximately $34 conversion factor, then translated back into dollars when paid under the PFS.

This preserves the approximate payment level while integrating the code into the PFS structure. However, ordinary PFS geographic adjustments and future conversion-factor changes mean that actual payments need not exactly match the OPPS amount in every location or future year.

 

Other consequences

Like the OPPS rule, the PFS final rule confirms that these services are other diagnostic tests rather than CLFS laboratory services. Consequently, conventional PFS beneficiary cost-sharing will apply, and CMS confirms that the CLFS date-of-service rule will not.

CAP, in its roles supporting CLIA and lab quality, has certainly viewed algorithms and  AI and machine learning as part of its modern domain.   Will CMS and CLIA central office weigh in on that/

The final rule responds to concerns about whether independent laboratories can bill directly for these services primarily by stating that the codes will be actively priced under the PFS. To my surprise, CMS acts dumb and does not fully work through the operational implications for laboratories performing computational analyses independently of physician offices or hospital outpatient departments.

CAP has long suggested CMS put genetic tests under RVUs (dating back to 2013).  Algorithm lab codse will be priced via RVUs, so CAP may revive its proposal to match.

The same important question about new test valuation remains. For the first 14 codes, CMS had existing CLFS prices to anchor the OPPS rates, which in turn anchor the PFS RVUs. That historical reference will not necessarily exist for future codes. The final rule does not fully resolve how new services will obtain an initial valuation (the one that is crosswalked into NT APCs) outside the established CLFS process.

3. OPPS General SaMS Final Rule

See CMS-1850-FC, Section X.B.1, preceding the OPPS laboratory discussion

This broader policy is important because the laboratory changes are an extension of CMS's effort to establish a consistent Medicare payment category for algorithm-based diagnostic software.

From SaaS to SaMS

CMS finalizes the terminology change from Software as a Service (SaaS) to Software as a Medical Service (SaMS).

The category encompasses algorithm-driven services supporting medical decision-making, including analysis of imaging, cardiovascular information, and other previously acquired clinical data.

The original proposal identified 36 SaMS codes, including services already receiving separate payment under clinical APCs or new technology APCs.

Standardization through new technology APCs

CMS generally finalizes reassignment of separately paid SaMS services from clinical APCs to new technology APCs, while maintaining payment levels reasonably close to CY 2026 amounts.

CMS also adopts status indicator O1, with the same payment specifications as status indicator S. Thus, O1 services are separately payable and are not subject to multiple-procedure discounting.

However, the general SaMS policy does not make every algorithm-based service separately payable. CMS preserves existing conditional packaging for certain Q1 services and complete packaging for certain N services. For example, it declines requests to change the conditional packaging status of retinal imaging codes 92227 and 92228.

These distinctions are important when considering how CMS may eventually structure payment for laboratory-derived SaMS. 

FDA authorization, coverage, and specific code decisions

For the general SaMS services in Section X.B.1, CMS requires FDA marketing authorization for Medicare payment. This contrasts with the laboratory-derived SaMS category in Section X.B.2, where such authorization is not currently required.

CMS also emphasizes that SaMS classification and APC assignment are payment decisions, not guarantees of Medicare coverage or medical necessity.

Several individual code decisions illustrate the interim approach. CMS added Cardiosense code 1036T at $150.50, burn-analysis code 0972T at $250.50, and EEG-analysis code 1041T at $150.50, resisting some requests for substantially higher initial payment. In other instances, it retained the proposed APC despite industry arguments that existing resource costs justified an increase.

Longer-term SaMS policy remains unsettled

The general SaMS discussion is explicit that 2027 represents an interim step, not a permanent payment framework. CMS received comments advocating dedicated cost centers and revenue codes, better reporting of software licensing and implementation costs, and payment approaches incorporating clinical value.

CMS indicates that it will consider these suggestions in developing possible CY 2028 proposals.

This creates a central uncertainty for both imaging-based and laboratory-based SaMS: whether the eventual permanent policy will retain individual separate payments, introduce more systematic cost-based valuation, adopt different payment categories, or increase packaging.

Overall significance

The immediate effect of the 2027 rules is principally a change in Medicare payment classification, rather than a wholesale repricing of existing tests.

But the longer-term implications are potentially much larger. By removing stand-alone computational analyses from the CLFS, CMS moves these services away from laboratory-specific pricing and billing protections and into the OPPS/PFS structures, with their coinsurance, budget-neutrality, geographic-adjustment, and packaging mechanisms.

The unresolved challenge is particularly striking for new tests: CMS has established where they will be paid, but has not yet fully established how their initial payment amounts will be determined. That may become one of the most consequential questions for algorithm-driven genomics and molecular diagnostics over the next several rulemaking cycles.

APPENDIX 1 - BUNDLING/PACKAGING

The distinction is that S is a general-purpose separate-payment indicator, while O1 is a new, software-specific separate-payment indicator. Both provide separate APC payment under OPPS, but O1 identifies Software as a Medical Service (SaMS) and, under CMS's interim 2027 policy, generally uses New Technology APCs rather than conventional clinical APCs.

A Hidden Advantage for Algorithmic Services Under OPPS. Moving standalone algorithms off the Clinical Laboratory Fee Schedule (CLFS) could improve their hospital outpatient reimbursement. Most conventional laboratory tests carry status indicator Q4, meaning payment is generally bundled into an accompanying clinic visit, emergency department visit, or other outpatient service. By contrast, CMS has established a new status indicator, O1, for Software as a Medical Service (SaMS), providing separate payment through an APC. O1 resembles the existing S indicator, which also provides separate payment without multiple-procedure discounting, as exemplified by autonomous retinal diagnostic code 92229. The difference is that O1 specifically identifies software-based medical services and assigns them to New Technology APCs under an interim payment policy, rather than treating them as conventional clinical procedures. Although CMS may revise the methodology, the immediate implication is significant: moving an algorithm out of the laboratory category can replace routine Q4 packaging with a pathway to separate hospital outpatient payment.

APPENDIX 2 - -A HARDWARE / WETWARE SAMS PARADOX 

###

One clear paradox.  CMS states that algorithmic lab services with no wetlab component are not CLFS, but if the algorithmic lab service DOES contain a real-world (wet lab0 component, it isn't SAMS and stays on PFS.    In sharp contrast, SAMS in general can contain both real-world and algorithmic components, like 92229 (digital camera, lens, focus, lighting, etc PLUS AI algorithm to render retinal diagnosis w/o physician work. OPPS table 84. [BQ]

### [AI]

The SaMS Paradox: When Physical Work Counts—and When It Doesn't

There is a striking inconsistency in CMS's reasoning, although the agency has left itself a possible legal distinction.

For laboratory-derived SaMS, CMS draws a fairly sharp line: if a CPT code describes only secondary computational analysis of previously generated laboratory data, it is no longer a clinical laboratory test and leaves the CLFS. If the code encompasses actual specimen analysis, as with 0511U or 81416, CMS agrees it is an integrated laboratory service and leaves it on the CLFS.

Yet SaMS generally is not restricted to software-only services.

Consider CPT 92229, which CMS retains as SaMS in OPPS Table 84. The service encompasses retinal imaging and autonomous algorithmic interpretation. The camera, optics, illumination, focusing, and image acquisition are integral to obtaining the diagnostic result. It is hardly a service delivered by an algorithm alone. Nevertheless, CMS assigns the entire service to SaMS, with status indicator O1 and new technology APC 1502.

There are even more explicit examples. CPT 1105T includes actual acquisition of EEG and ECG signals, followed by algorithmic analysis. CMS newly designates it SaMS. Similarly, CPT 0972T includes multispectral imaging, system setup, and image acquisition, yet also qualifies as SaMS.

Thus, CMS is not consistently applying a principle that SaMS must consist exclusively of computational work.

Can the distinction be defended?

CMS could argue that it is answering two different questions.

In the laboratory context, the question is whether a service qualifies under the statutory clinical diagnostic laboratory test benefit category, with its particular payment and regulatory requirements. CMS treats secondary analysis of previously obtained data as an other diagnostic test.  But if wetlab is done, that overrides SAMS and "pulls" the test into CLFS.

In the broader SaMS context, the question is whether algorithm-driven diagnostic technology belongs in the new SaMS payment category, even when the service also includes the physical acquisition of the underlying data.  (See retinal camera 92229 and its digital AI report, below).  

That is a plausible legal distinction. But it is not a consistent functional definition of SaMS.

Indeed, the distinction becomes particularly awkward for pathology-based software that includes preparation or digitization of glass slides. Codes 1063T and 1064T explicitly contemplate H&E staining and slide digitization when performed, yet CMS includes them among the 14 laboratory-derived SaMS services leaving the CLFS.  So CMS is treating these as if the wetlab doesn't exist, and presumably doesn't value it.  But CPT doesn't want you to bill the wetlab service (e.g. IHC) if the  CPT code absorbs it, "if performed."   So you're blocked from billing for example IHC x 7, if performed.

So the presence of physical work is disqualifying in some laboratory codes, permissible in other laboratory codes, and entirely compatible with SaMS designation in imaging and other diagnostic specialties.

The most charitable interpretation is that CMS is trying to preserve existing integrated laboratory tests while carving out independently identifiable computational services. However, the resulting boundary depends heavily on the individual CPT descriptor, its historical coding conventions, and CMS's interpretation of optional physical components.

That is a precarious foundation for an increasingly consequential payment classification.  CMS does not discuss that wetlab boots you out of SaMS in lab and lab data world, but physical events (like the 92229 physical camera) is compatible with treatment as a clinical software service.

And the paradox has a series of practical reimbursement consequences. Whether physical activity is included in a single CPT code, or separated into two codes describing acquisition and subsequent analysis, may determine whether a service remains on the CLFS, receives conventional coinsurance, escapes the laboratory date-of-service rule, and ultimately becomes subject to OPPS packaging.

The larger lesson may be that CMS has created a payment category called Software as a Medical Service without yet establishing any consistent principle for how much non-software activity that service can contain. The agency acknowledges that its 2027 policy is interim. The unresolved definitional problem may prove considerably harder to resolve than the initial payment assignments.

###

The paradox is not merely semantic. CMS applies opposite reasoning to the presence of physical diagnostic work, depending on whether the service originated in the laboratory or elsewhere.

Consider the contrast:

- CPT 92229 (retinal imaging): Physical image acquisition, involving a camera, optics, illumination, and positioning, is combined with an algorithmic diagnostic interpretation. CMS classifies the entire service as SaMS.

- CPT 0511U (tumor drug-response testing): Physical laboratory activity is combined with computational prediction. CMS concludes that the physical laboratory component prevents the service from being classified as SaMS, retaining it on the CLFS.

Both are integrated diagnostic services involving physical procedures and algorithms. Yet in one case the physical component is compatible with SaMS classification; in the other it is grounds for exclusion from a lab SaMS.

CMS can invoke the distinct statutory treatment of laboratory tests to explain the administrative result. But that does not resolve the conceptual inconsistency in what constitutes Software as a Medical Service.

The paradox is that physical diagnostic work does not disqualify a service from being SaMS—unless that physical work happens to be laboratory work.

And that makes SaMS less a coherent technological category than a classification shaped by the historical boundaries of Medicare payment systems.

There is no need to force a reconciliation. The inconsistency is itself the important finding about these tricky final rules.

###

83861 tears

 ophthalmologists microfluidic analysis of tears, 83861.

6.2M out of 7M.

PAMA

22.48 to 22.03

Cologuard Quotations about COGS of Cologuard Plus

 Direct quotations from Chat GPT including prompt.

##

   Let's pivot back to Cologuard coding.   Cologuard Plus 0464U was priced by CMS in CY2024, inititially proposed at parity to Cologuard Original about $520 and   later raised by cMS to 81327x3 or nearly $600.   

However, I seem to recall statements by Exact sciences that the operating costs or COGS or Cologuard Plus would be less - COGS is less - than had been the case for Cologuard Original.   Your assignment is to use agentic reserach and search the internet for quotations or statements where Exact sciencee (or, if demarcated as such, analysts) imply or statte that COGS goes down for Cologuard Plus.    

###

Yes—your recollection is correct, and the evidence is explicit and repeated. 

Exact Sciences management described the next-generation Cologuard test as costing at least 5% less per test in 2023, reiterated lower COGS during the 2024 pricing process, and stated in May 2025 that Cologuard Plus cost about 5%–6% less to run than original Cologuard. These are company statements, not merely analyst inferences. The Motley Fool

• There is also a useful rediscovery: you covered this yourself on November 6, 2024, in “Exact Sciences—Earnings Call a Tutorial in Crosswalk and ADLT pricing.” Your post specifically highlighted the lower cost of goods for Cologuard Plus. discoveriesinhealthpolicy.com

The strongest passages follow. Quotations are short exact excerpts; the surrounding explanations are paraphrases.

1. June 21, 2023: Company-hosted Investor Day transcript—explicit COGS reduction and an operational explanation

Jeff Elliott, CFO, discussing the next-generation product then called Cologuard 2, said:

“we expect at least a 5% reduction in cost of goods per test”

In the preceding answer, Jake Orville explained that one of the two laboratory processing lines described during the investor tour would go away with Cologuard 2, connecting that simplification to approximately 5% lower costs.

Elliott placed the product improvement within a broader ambition to reduce Cologuard cost per test from approximately $125 to $100 or less. The entire $125-to-$100 reduction should not be attributed to Plus alone: he also discussed continued improvements and laboratory operating leverage. s22.q4cdn.com

Exact Sciences Investor Day transcript, PDF pages 29–30

This is particularly useful evidence because the transcript is hosted with the company’s investor materials and predates the CMS pricing proceedings.

2. August 1, 2023: Q2 earnings call—direct comparison with original Cologuard

Jeff Elliott, CFO and COO, answering a question about Cologuard 2 cost of goods:

“the cost per test of Cologuard 2 will be at least 5% lower than what it is today for Cologuard 1.”

He attributed the reduction to the development of more accurate markers, automation, and efficiencies incorporated into the new test. This is the clearest early statement comparing the two generations directly. The Motley Fool

Q2 2023 earnings-call transcript

3. March 14, 2024: Company presentation—fewer markers and better laboratory throughput

Slide 6 of Exact Sciences’ BLUE-C presentation states:

“Cologuard Plus delivers performance and cost advantages”

The slide depicts a reduction from 11 markers to five and identifies improved laboratory throughput. It does not quantify COGS savings, but it provides company-authored support for the operational explanation behind the earnings-call statements. s22.q4cdn.com

Exact Sciences BLUE-C presentation, slide 6

4. November 5, 2024: Q3 earnings call—higher selling price and lower COGS discussed together

Aaron Bloomer, CFO, described an expected gross-margin benefit in 2025:

“with the launch of Cologuard Plus, which will give us added ASP as well as lower cost of goods.”

This statement occurred during the same fall in which CMS was finalizing the new test’s payment determination. It explicitly identifies two separate favorable effects: higher average selling price and lower production costs. The Motley Fool

Q3 2024 earnings-call transcript

5. May 1, 2025: Q1 earnings call—approximately 5%–6% lower cost to run Plus

Aaron Bloomer, CFO, answering Barclays analyst Luke Sergott’s question about gross margins:

“the actual cost to run the test is lower by about 5% or 6% than it is for Cologuard.”

The immediately preceding wording identifies Cologuard Plus as the test being compared with original Cologuard. Bloomer also acknowledged potential transition costs from sorting specimens and running the two product lines concurrently. He expected the pricing benefit to more than offset those potential headwinds. marketbeat.com

Q1 2025 transcript, approximately 44:25 into the call

This is the strongest later confirmation: the company was still describing lower unit operating costs as the product entered commercial use.

Secondary reporting also noticed the combination of higher requested payment and lower costs.

• July 26, 2024—Investing.com: Its report on the Medicare advisory-panel discussion juxtaposed Exact’s requested 25% price increase with the company’s expectation of 5% lower COGS. This is secondary financial reporting, not a CMS finding or an original analyst research note. The page discloses AI assistance and editorial review. investing.com
• November 26, 2024—Investing.com, reporting on Leerink Partners’ outlook: The article described the favorable final Medicare pricing decision alongside 5% lower COGS for Plus, attributing that cost expectation to company management. Again, this is a media account of the analyst outlook, rather than the underlying Leerink note. investing.com

For your purposes, the direct company statements above are stronger evidence than either secondary article.

The defensible conclusion is that Exact repeatedly represented Plus as a less expensive test to produce, despite seeking and obtaining higher reimbursement. The evidence concerns per-test COGS or the cost of running the assay. It does not mean total corporate spending fell, nor does it independently establish an audited, fully allocated cost for each product.

A concise formulation you can use is:

Exact Sciences repeatedly told investors that Cologuard Plus would have lower per-test costs than original Cologuard. Management projected at least a 5% reduction in 2023, reiterated lower COGS alongside anticipated higher selling prices in November 2024, and described a 5%–6% lower cost to run the test in May 2025. The Motley Fool

One small correction to the coding premise: CMS’s official CY2025 mapping is 0464U → 81327 × 3 + 82274—81327, rather than 83127 [i fixed in my prompt above], and including the fecal hemoglobin component. CMS documented that crosswalk in its December 13, 2024 transmittal. cms.gov

##

private link at

https://chatgpt.com/c/6ad998cf-d4a4-83e8-9333-9c882d6c5a47

Pulling FDA Docket Comments per Chat GPT

 Yes—you can automate both the downloading and the screening, without manually opening 129 PDFs. And I found a particularly useful shortcut for this exact docket.

Someone has already collected and analyzed much of it. Innolitics published a “Running Census” of FDA-2026-N-7874. Its author reports retrieving 107 comments and 92 PDF/Word attachments through the Regulations.gov API, then extracting their text. The version I could read describes a September 21 snapshot, so it should not be treated as complete coverage of the 129 you see now. Nevertheless, it provides an immediate starting point for finding substantive comments. Innolitics

For your own complete collection, here are the practical options:

What you want	How to accomplish it
Download everything without clicking each comment	Use the official API to enumerate comments and retrieve each comment’s attachment links; a script downloads the files automatically.
Find comments of three or more pages	Have software count pages in every downloaded PDF and create a sortable index. No manual opening required.
Identify organizations hidden behind personal names	Extract the attachment text and examine letterheads, opening paragraphs, and signatures. Preserve the portal’s submitter name in a separate column.
Search reliably for “California”	Search the locally extracted text of both comment bodies and attachments, using OCR for scanned files.
Prioritize substantive submissions	Combine page count, word count, organizational affiliation, and the presence of concrete recommendations.

The official Regulations.gov API documentation supports retrieving comment details with attachments and explains API-key registration. This is an automated sequence of downloads, rather than necessarily a single “download docket ZIP” button. GSA Open Technology

Your “California” experience is a reason not to trust a zero-result search as proof of absence. One plausible explanation is that the searchable listing contains “John Smith” and “See attached,” while “University of California Health” appears inside the PDF. Attachment indexing, scanning, or posting delays could also matter. I haven’t verified which explanation applies to that particular submission, so I would not categorically claim that the portal never searches attachments.

There is also an Export route, with an important distinction: exporting a spreadsheet of comment records is not the same as downloading their attachments. George Washington University’s Regulatory Studies Center documents an export-plus-script workflow and provides public code for downloading comments and converting attachments to text. Its instructions refer to an older interface/API, so I would use it as a model rather than assume it runs unchanged today. I could not verify the current export-button placement through my web access. GitHub

For your research, the most useful result would be a comment index with:

• Organization identified from the actual letter.
• Portal submitter name and comment ID.
• PDF page count and extracted word count.
• A one-sentence description of the argument.
• Direct attachment link.
• Flags for “three-plus pages,” “needs OCR,” and “duplicate.”

I would use three-plus pages as a priority flag, not an exclusion rule: a two-page institutional letter can contain an important position, while a long attachment may mostly reproduce background material. The automated approach makes both visible.

#### HOW DO IT

####

Absolutely. My previous answer skipped over the difficult part: “use the API” is an instruction for a programmer, not a usable set of directions.

And you’re right that we need the comments submitted near October 19. The September snapshot is insufficient for that purpose. I haven’t verified the submission-date breakdown, so I can’t confirm whether those late comments constitute the majority.

I found an easier place to start: Regulations.gov has a “Bulk Data Download” feature that does not require programming. GSA documents it specifically as a way to download information without technical expertise. It emails you a spreadsheet file. GSA

Here is the plain-English procedure:

1. Open Regulations.gov.
2. Scroll to the bottom of the page and look for “Bulk Data Download.” Alternatively, look on the homepage for “Create Custom Download.”
3. Choose the option to download by Docket ID.
4. Paste this identifier: FDA-2026-N-7874
5. Follow the prompts to request the download and provide your email address.
6. Look for an email from no-reply@regulations.gov. Download the CSV file using its link. “CSV” simply means a spreadsheet that Excel can open.
7. Upload that file here. You don’t need to clean it up first.

Those are GSA’s published directions; the announcement is archived, so the precise button wording or placement may have changed. If your screen differs, send me a screenshot. GSA

Important: that first download is the catalog of comments—not a folder containing all the PDFs. Once you upload it, I can inspect whether it includes the attachment download links. That determines the simplest way to collect the actual files in a batch.

For comparison, here is what the API approach means in ordinary language:

Term	What it means
API	A way for a program to ask Regulations.gov for information automatically.
API key	A registration code that identifies the program’s requests.
Download script	A small program that asks for the comment list, finds each attachment, and downloads the files for you.

The API route requires registering for a key, then running a prepared program. Getting a key alone does not produce the PDFs. The official API supplies comment details and attachment information; the program handles the repeated downloading. GSA Open Technology

For you, I suggest starting with the seven steps above. The immediate goal is to get the current docket’s spreadsheet here. Then we can work toward a folder of PDFs and an index showing organization, page count, and subject—without your opening each submission yourself.

Four Transcripts and a Funeral: AI Reads Invitae's Last Year

 (We asked AI to read the last 4 OPKO investor calls prior to the divestiture of Bioreference to Labcorp.  Here, we show the result when we gave the same task but for Invitae's last 4 investor calls.)

 ###

Ernst & Young gave the 2022 financials an unqualified audit opinion without a going-concern paragraph. The formal “substantial doubt” conclusion first appears in management’s Q3 2023 10-Q; Invitae then highlighted it explicitly on the November earnings call. SEC

Here is the first-pass case study.

Invitae: The Last Four Calls

A business case in growth, capital structure, genomic medicine, and the narrowing path from turnaround to Chapter 11

Invitae Corporation, February 2023–February 2024

Executive Summary

Invitae entered 2023 describing a dramatic corporate restructuring as largely accomplished. Management had cut costs, sold noncore assets, refinanced near-term debt, and projected a cash runway through the end of 2024. Across its next two quarterly calls, the company repeatedly emphasized improving gross margins, sharply reduced cash burn, a “strong cash position,” and approximately $250 million of additional secured borrowing capacity. By August, Invitae was actually lowering its projected 2023 cash burn despite reducing revenue guidance. Three months later, however, the language changed abruptly: the Q3 10-Q stated that substantial doubt existed about Invitae’s ability to continue as a going concern, while a board special committee began evaluating capital raises, debt restructuring and asset sales. Ninety-seven days after that final earnings call, Invitae filed Chapter 11. Labcorp ultimately acquired the operating business for $239 million. The case illustrates a critical distinction: operational improvement can be real while the capital structure has already become untenable.

I. Background: Building the Genetics Platform

Invitae was founded around a compelling and unusually ambitious proposition: genetic information should become inexpensive enough, comprehensive enough, and sufficiently integrated into ordinary medicine that genomic testing would move from a specialist service into mainstream healthcare.

The strategy required scale. Invitae drove testing prices downward, expanded its menu, accumulated genetic data, built relationships with biopharmaceutical companies, and pursued acquisitions intended to create a comprehensive genetics platform.

For a period, capital markets rewarded the ambition.

The acquisition program eventually extended well beyond conventional hereditary-disease germline testing. Invitae purchased pharmacogenomics capabilities, reproductive-health businesses, digital tools and oncology technologies. The largest transaction was the 2020 acquisition of ArcherDX. Invitae agreed to pay 30 million shares plus $325 million cash upfront, with as many as another 27 million shares contingent on milestones. At announcement, the transaction was valued at approximately $1.4 billion. Management described the combination as creating a global leader spanning inherited cancer risk, tumor profiling, liquid biopsy and cancer monitoring.

The strategy produced breadth, but it also produced complexity, spending and substantial capital requirements. By 2022 the environment had changed. Growth-oriented biotechnology and diagnostics companies could no longer assume inexpensive capital would remain indefinitely available.

In July 2022, Invitae changed direction dramatically. Founder-CEO Sean George was succeeded by COO Ken Knight. The company announced a strategic realignment that included eliminating more than 1,000 positions, consolidating facilities, abandoning noncore products and reducing its international presence from more than 100 countries and territories to fewer than a dozen. Management estimated approximately $326 million in annualized savings once the restructuring was fully implemented.

The bankruptcy court record would later characterize the period even more starkly. Invitae experienced turnover among four CFOs over approximately two years, alongside other C-suite changes. In the company's own first-day bankruptcy declaration, that turnover was described as having delayed the company's response to its challenges and impeded development of a cohesive go-forward strategy.

Thus the four final earnings calls did not begin with a healthy growth company suddenly entering distress. They began midway through an attempted rescue.

II. The Four Calls

There was no Q4 2023 earnings call. Invitae's Q3 2023 call on November 8, 2023 became its final quarterly investor call. The natural four-call sequence therefore begins with Q4 2022 and ends with Q3 2023.

Quarter	Call date	Public financial narrative
Q4 2022	February 28, 2023	Debt problem substantially addressed; runway through 2024
Q1 2023	May 9, 2023	Strong execution; burn falling faster than expected
Q2 2023	August 8, 2023	Revenue softens, but burn forecast improves further
Q3 2023	November 8, 2023	“Substantial doubt”; special committee; capital structure options
Chapter 11	February 13, 2024	Court-supervised sale process

February 28, 2023 — Q4 2022: “A More Stable Financing Footing”

This was the optimistic opening chapter.

The 2022 restructuring appeared to be working. Invitae reported approximately $122 million of fourth-quarter revenue. Ongoing quarterly cash burn had fallen to about $77 million, compared with $196 million a year earlier. Management projected more than $500 million in 2023 revenue and full-year ongoing cash burn of $250–275 million — less than half the 2022 level.

But the principal strategic message concerned the balance sheet.

Invitae had arranged a $336 million transaction led by Deerfield that exchanged most of the convertible debt due in 2024 for new secured notes extending into 2028. It also repaid a $135 million term loan. CFO Roxi Wen said the transactions would reduce total debt by more than $165 million and shrink the 2024 maturity from approximately $485 million to only $14 million.

Management estimated a pro forma cash balance of approximately $450 million.

The conclusion was emphatic: Invitae expected its cash runway to extend through the end of 2024 and said it retained approximately $245 million of additional secured debt capacity. Wen characterized the refinancing as a “fundamental change” benefiting Invitae's capital structure and financial health.

CEO Ken Knight's operating message was similarly constructive. Invitae had “stabilized” its portfolio. Profitable growth was now the foundation of the 2023 business plan, while investment could continue in new growth opportunities such as personalized cancer monitoring.

Case interpretation: February 2023 did not sound like the beginning of a bankruptcy countdown. It sounded like the end of a financial crisis. Management presented both halves of the turnaround — operations and financing — as materially improved.

There was, however, a less comforting fact beneath the refinancing. The new 2028 notes were secured by a first-priority lien on substantially all of Invitae's and its subsidiaries' assets, including intellectual property. The company had bought time, but the capital structure was becoming increasingly creditor-centered.

III. May 9, 2023 — Q1: “Same Mission, New Path”

The first-quarter call reinforced the turnaround narrative.

Revenue was $117.4 million. Non-GAAP gross margin reached 47.9%, compared with 36.6% for 2022. Ongoing cash burn was approximately $51 million, about 70% below the prior-year Q1. Invitae reported approximately $389 million of cash, restricted cash and marketable securities after repaying the term loan.

Management again referred to its “strong cash position” and emphasized nearly $250 million of available secured debt capacity. Revenue guidance remained above $500 million for 2023. Cash-burn guidance remained $250–275 million, with management saying results were tracking toward the favorable end of the range.

Knight ended the prepared remarks with language that now reads strikingly in retrospect: the near-term debt had been “substantially” addressed, the balance sheet had improved, and additional secured capacity remained available.

His summary was: “same mission, new path.”

The contemporaneous 10-Q supported the public presentation, although with important caveats. Invitae stated that its cash, investments and operating receipts should be sufficient for at least the following twelve months. It simultaneously acknowledged that it expected to require additional financing before profitability and that such financing might not be available on acceptable terms.

Case interpretation: Q1 may be the point at which a reader looking only at quarterly operating trends could most easily conclude that Invitae had escaped its crisis. Cash burn was collapsing; margin was improving; immediate debt maturities had been pushed out.

The unresolved issue was not yet whether the underlying laboratory business could improve. It clearly was improving.

The question was whether it could improve fast enough to outrun the balance sheet.

IV. August 8, 2023 — Q2: Better Cash Burn, Worse Revenue

The second-quarter call contains perhaps the most instructive contradiction in the entire case.

Knight opened by saying Invitae was continuing its “steady march toward becoming a profitable business.” Non-GAAP gross margin reached 49.8%, its eighth consecutive quarter of improvement. Ongoing cash burn was approximately $53 million, 64% lower than the comparable quarter in 2022.

Cash and investments stood at approximately $336 million, down from $389 million at the end of Q1. Again, management described the balance as a “strong cash position.” Again, it said approximately $250 million of additional secured debt capacity remained available.

There were genuine operating problems. Revenue was approximately $121 million, down from $137 million reported in the prior-year period, although management emphasized that discontinued operations distorted the comparison. On a pro forma basis, growth was only about 1%.

Hereditary-cancer reimbursement had weakened. Oncology fee-for-service revenue was soft. Invitae abandoned pharmacogenomic testing because the reimbursement path and required investment no longer fit management's desired time horizon.

Consequently, management reduced 2023 revenue guidance from more than $500 million to $480–500 million.

Ordinarily, a revenue-guidance cut at a cash-burning company might imply deterioration in liquidity expectations.

Invitae reported the opposite.

Management improved its cash-burn guidance from $250–275 million to $220–245 million, saying the company was ahead of plan.

Knight concluded that one year after the strategic realignment, Invitae had better visibility into the portfolio and had created “doable paths for profitable growth tomorrow.”

The Q2 10-Q still concluded that Invitae's available liquidity would cover at least the next twelve months, although it explicitly stated that external capital would be needed beyond that period and that such funding had not been secured.

Case interpretation: August 2023 is the most deceptive moment if operating metrics are viewed independently of financing conditions.

Gross margin was improving. Cash burn was substantially lower. Burn guidance was getting better.

Yet the company's dependence on future financing had not disappeared.

Invitae had solved a rate-of-burn problem more successfully than it had solved a capital-structure problem.

V. November 8, 2023 — Q3: The Language Changes

Three months later, the central premise of the story changed.

Operationally, Invitae could still report substantial progress. Q3 revenue was $121.2 million, approximately 4% higher on management's pro forma basis. Non-GAAP gross margin reached 52.4%, the ninth consecutive quarter of expansion. Management reaffirmed its $480–500 million revenue forecast and $220–245 million ongoing cash-burn guidance.

If the case consisted only of revenue, margin and operating-efficiency slides, the turnaround still looked alive.

Then Knight said:

“Today you will see going concern language in our 10-Q.”

The accompanying filing was unequivocal. Because of losses, projected cash needs and existing liquidity, “substantial doubt exists about the Company's ability to continue as a going concern.” Management said continued operation depended on successfully reducing costs and improving liquidity. Alternatives included new debt or equity, asset sales, business and R&D refocusing, and further reductions in spending. If adequate funding could not be obtained, the filing specifically identified liquidation or bankruptcy proceedings among the possible outcomes.

The board had formed a special committee focused on the capital structure. Management told investors it was exploring raising capital, addressing debt, selling assets and further cost reductions.

New CFO Ana Schrank reported approximately $265 million of cash, restricted cash and marketable securities. Quarterly ongoing cash burn had increased to approximately $64 million, partly because of interest and other one-time payments. She emphasized that Invitae still had sufficient liquidity to operate normally while management worked with stakeholders on a path forward.

The company continued discussing 2024.

Management talked about future margin expansion, product mix and continued growth. Knight said Invitae remained “pretty darn confident” that gross-margin improvement could continue into the future.

Case interpretation: Q3 contains two narratives simultaneously.

The operating business was demonstrably becoming more efficient.

The corporation financing that business was running out of strategic freedom.

That distinction is the heart of the Invitae case.

SIDEBAR: The Going-Concern Threshold

The accounting progression is worth isolating because hindsight can otherwise make bankruptcy appear more obviously imminent than it actually looked in contemporaneous filings.

At December 31, 2022, Invitae stated that available cash and investments should cover at least twelve months. Ernst & Young issued an unqualified audit opinion on the 2022 financial statements and did not add a going-concern explanatory paragraph.

At March 31, 2023, management again stated that existing liquidity should satisfy expected cash needs for at least twelve months, while acknowledging that more capital would ultimately be required.

At June 30, 2023, management again made essentially the same twelve-month liquidity determination.

At September 30, 2023, the conclusion flipped. Management determined that losses, future cash requirements and available liquidity raised substantial doubt about Invitae's ability to continue as a going concern.

So the important event was not a long-standing auditor qualification that investors ignored. It was an abrupt transition in management's own required ASC 205-40 assessment between the Q2 and Q3 filings.

That makes the November call more consequential, not less.

The company crossed a formal accounting threshold at almost exactly the same moment that its public capital-allocation discussion shifted from financing a turnaround toward restructuring the enterprise itself.

VI. Ninety-Seven Days Later

On February 13, 2024, 97 days after the Q3 call, Invitae and several subsidiaries filed voluntary Chapter 11 petitions in New Jersey.

The filing was not designed simply to reorganize debt and return the same public corporation to shareholders. Invitae entered bankruptcy with a transaction-support agreement backed by holders representing approximately 78% of its secured notes and pursued a court-supervised sale of the operating business.

The contemplated plan provided that existing equity interests would be cancelled without distribution.

The debt refinancing celebrated one year earlier now assumed a different significance. The secured notes issued in 2023 had claims against substantially all of Invitae's assets. In bankruptcy, their holders were among the parties supporting the sale process.

This does not make the February 2023 refinancing irrational. Without it, Invitae faced approximately $485 million of 2024 maturities and might have failed sooner. The transaction bought time.

But it bought time, not solvency.

VII. Labcorp: $239 Million for the Operating Platform

The bankruptcy auction eventually produced a familiar buyer: Labcorp.

Labcorp won the auction for selected Invitae assets and agreed to pay $239 million in cash, plus assumption of specified liabilities and other noncash consideration. The bankruptcy court approved the transaction in May 2024.

Labcorp expected the acquired operation to generate approximately $275–300 million of annual revenue, principally in specialty genetics such as oncology and rare disease.

That relationship between revenue and purchase price is striking.

Invitae had once assembled a broad genetics platform through years of investment and acquisitions. ArcherDX alone had carried an announced transaction value of approximately $1.4 billion.

The surviving operating genetics platform was ultimately sold out of bankruptcy to Labcorp for $239 million.

Those numbers should not be treated as a clean investment-return calculation. The businesses, assets, liabilities, market conditions and time periods are different. But as a corporate-strategy picture, the contrast is extraordinary.

VIII. What Actually Failed?

The easiest explanation is that Invitae's business failed.

That is too simple.

During the terminal four calls, several elements of the operating business were moving in the right direction. Gross margins rose for nine consecutive quarters. Cash burn fell dramatically from its 2021–2022 levels. Certain hereditary disease and women's-health businesses continued growing. The company accumulated one of the industry's largest clinical genetics datasets and retained genuine technological capabilities.

Labcorp's willingness to pay $239 million and continue the business confirms that the operating assets were not worthless.

A more useful diagnosis is that Invitae's capital structure failed before the operating turnaround could reach self-sustainability.

The company had built its platform during an era when equity and debt capital were unusually available. It made acquisitions, expanded globally and tolerated very large operating losses in pursuit of scale.

When capital markets changed, management attempted to reverse that model rapidly: fewer countries, fewer employees, fewer products, higher-quality revenue, better reimbursement, lower costs.

The restructuring achieved impressive operating savings.

But a company burning $50–60 million per quarter while carrying substantial secured and convertible debt does not merely need improving margins. It needs either positive cash flow soon enough or continuing access to outside capital.

By Q3 2023, the latter assumption had broken.

The 10-Q explicitly referred to difficult debt markets. The board was no longer merely optimizing operations. It was evaluating the capital structure itself.

IX. The Most Interesting Feature of the Four Calls

The most revealing sequence is therefore not:

Optimism → deterioration → bankruptcy.

It is:

Operational deterioration → restructuring → genuine operational improvement → financing failure → bankruptcy.

That is a considerably more interesting business case.

In February 2023 management could accurately say that cash burn had improved dramatically and near-term maturities had been pushed out.

In May, it could accurately say that gross margins were improving and burn was ahead of plan.

In August, it could accurately say that cash-burn guidance had improved further.

And in November, it could accurately report its best gross margin in years while simultaneously concluding that substantial doubt existed about the company's ability to continue.

All four statements can be true.

That is precisely what makes the case valuable.

X. BioReference and Invitae: An Unexpected Labcorp Parallel

The coincidence with BioReference is worth noting.

Labcorp ultimately acquired both businesses, but through radically different paths.

BioReference spent its terminal year inside OPKO being rationalized, narrowed and publicly described as approaching sustainable profitability. OPKO then negotiated a strategic sale at a substantial price.

Invitae spent its terminal year being rationalized, narrowed and publicly described as moving toward profitable growth. But its capital structure deprived management of the luxury of waiting for that thesis to mature. The sale occurred only after Chapter 11, under court supervision, for $239 million.

In both cases, scale mattered.

A testing platform that struggled to create adequate returns as an independent or noncore enterprise could be more valuable inside Labcorp, where laboratory infrastructure, contracting, logistics, payer relationships and commercial reach already existed.

The difference was bargaining power.

OPKO could sell BioReference from a position in which it still had strategic alternatives.

Invitae entered the sale process because its alternatives had largely disappeared.

XI. Conclusion: The Difference Between Runway and Escape Velocity

The phrase that echoes through Invitae's final year is cash runway.

Management improved it. It refinanced debt. It cut spending. It sold businesses. It reduced employment. It expanded margins.

But runway is not the same thing as escape velocity.

Invitae entered 2023 believing that a dramatically lower burn rate, improved reimbursement and renewed growth could give the operating business sufficient time to become financially self-sustaining.

For much of the year, the reported operating metrics supported that thesis.

What changed was the availability of the bridge between the present and that future.

By November, outside capital could no longer be treated as an ordinary planning assumption. The Q3 going-concern disclosure effectively acknowledged that operational improvement alone might not arrive quickly enough.

Ninety-seven days later, Invitae entered Chapter 11.

The core genetics platform survived.

The public corporation did not.

And Labcorp, rather than Invitae's shareholders, became the owner positioned to test whether the long-promised economics of scale in clinical genetics could finally be realized.

The part I find most interesting is Q2 → Q3. In August, Invitae still said it had a “strong cash position,” roughly $250 million of additional secured debt capacity, and actually improved its burn guidance. In November, the secured borrowing capacity essentially disappears from the rhetorical foreground and is replaced by going concern, stakeholders, special committee, debt, asset sales. That is the inflection point I would visually emphasize if we turn this into the companion PDF to the BioReference report. The Motley Fool