To: NOTE TO FILE
From: Bruce Quinn, MD, PhD
Date: August 28, 2026
Re: A Priority Issue for CAP Engagement—CMS, SaMS, CLIA, and Computational Pathology
Executive summary
What are current issues involving the College of American Pathologists that might warrant policy engagement.
AI recommends focusing on one unusually important and timely issue: CMS’s proposal to remove certain computational pathology tests from the clinical laboratory category and reclassify them as Software as a Medical Service, or SaMS.
- First, CAP has helped shape the coding architecture for digital and computational pathology through the AMA CPT process, including Category III digital pathology codes, pathology-specific algorithm codes, the evolving Clinically Meaningful Algorithmic Analyses framework, and CPT Appendix S for artificial intelligence.
- Second, CAP is the nation’s most influential real-world implementer of CLIA requirements. CLIA regulations change slowly; CAP’s accreditation checklists, committees, inspectors, and laboratory teams continually translate those regulations into workable standards for new technologies—including digital pathology, AI validation, remote interpretation, and distributed services involving more than one CLIA-certified laboratory.
CMS now proposes, in both the CY 2027 OPPS and PFS rules, that certain algorithmic analyses of laboratory data are not clinical laboratory tests at all. CMS would move them from the CLFS into the much broader statutory category of “other diagnostic tests” and pay them as SaMS. For computational pathology companies such as ArteraAI, Valar Labs, Paige, and others, the proposal could separate the billable algorithmic test from the CLIA laboratory framework in which CAP, CPT, developers, and laboratories have been placing it.
CMS is not formally declaring that every aspect of digital pathology lies outside CLIA. Its immediate proposal concerns stand-alone algorithmic analyses whose CPT descriptors do not name a conventional laboratory method. But the agency’s reasoning could reach directly into clinical computational pathology, where an algorithm examines an H&E whole-slide image and generates a new patient-specific diagnostic, prognostic, predictive, or treatment-selection result.
CAP has built leadership through CPT
CAP has been highly active in establishing the coding pathway for digital pathology. It worked with the AMA CPT Editorial Panel to create 13 Category III digital pathology add-on codes for 2023 and another 30 codes for 2024. CAP explicitly described those codes as an “on-ramp” for artificial intelligence and as a pathway toward eventual Category I recognition and national payment.[1]
That work fits within a wider CPT architecture for algorithmic services:
Multianalyte Assays with Algorithmic Analyses established that a laboratory test may combine laboratory measurements with an algorithm to produce a patient-specific clinical result.
Category III codes provide an entry point for new computational pathology tests and generate utilization data needed for later Category I consideration.
CPT Appendix S classifies AI-enabled medical services as assistive, augmentative, or autonomous. The AMA substantially revised Appendix S in 2026 to clarify what constitutes a clinically meaningful AI output.[2]
The AMA is developing a possible new framework, tentatively called Clinically Meaningful Algorithmic Analyses, for algorithms that produce medically actionable outputs even when no physician is directly involved at the point of service.[3]
CAP’s February 2026 comments to HHS described the AMA CPT Editorial Panel as a key external governing body for AI adoption. CAP urged HHS to work with CPT to create a clear and clinically appropriate coding framework for AI services.[4] CAP therefore sees coding not merely as billing mechanics, but as part of the governance and adoption pathway for digital pathology.
CAP has also built digital pathology leadership through CLIA accreditation
CAP’s second role is arguably even more important. CAP is a federally deemed CLIA accrediting organization, and a CAP inspection substitutes for a CMS inspection in CAP-accredited laboratories. CAP accredits more than 8,000 laboratories and updates its checklists annually, while the underlying CLIA regulations may remain substantially unchanged for years or decades.[5]
In practice, CAP continually builds the leading edge of “CLIA in real life.” This includes questions that the original regulations could not have anticipated:
How should a laboratory validate a whole-slide-imaging system?
Who is responsible for validating an AI algorithm before it is used for patient testing?
How should performance drift and software updates be monitored?
How should the laboratory maintain positive patient identification across scanners, image-management systems, algorithms, and reports?
What happens when slides are created at one CLIA laboratory, scanned at another site, analyzed by a third entity, and interpreted by a pathologist working under yet another CLIA certificate?
When is a remote site part of the primary laboratory, and when does it become a separate referral laboratory?
Which laboratory director is accountable for validation, quality management, records, and the final patient result?
CAP’s 2025 Laboratory General Checklist now contains a section specifically titled “Digital Pathology Including Remote Data Assessment.” CAP inspectors may review digital pathology policies, patient reports, validation records, image-matching procedures, scan-failure rates, image-quality criteria, discordance reconciliation, and risk-mitigation procedures. Checklist requirement GEN.50630 requires laboratory validation or verification and laboratory-director approval of digital pathology systems used for clinical purposes. GEN.52860 places digital pathology within the laboratory’s continuing quality-management system.[6]
CAP’s formal policy statements are equally clear. In February 2026, CAP told HHS that its accreditation checklists require laboratory directors to validate or verify new tests and methods before patient testing, expressly “including AI and machine learning algorithms.” CAP described FDA review and CLIA oversight as complementary and necessary, and stated that CLIA requires laboratory directors to assess AI systems before local implementation.[4]
In December 2025, CAP similarly told FDA that AI had become an important new element of the pathologist’s responsibilities as a CLIA laboratory director or section director. CAP emphasized that CLIA and CAP requirements extend beyond FDA authorization and manufacturer instructions to include local validation, controls, performance monitoring, corrective action, and management of changes that could affect patient results.[7]
The CMS proposal moves in the opposite direction from CAP
Section 1861(s)(3) of the Social Security Act contains three broad diagnostic categories reflecting the world of 1965:
Diagnostic X-ray tests;
Diagnostic laboratory tests; and
Other diagnostic tests.
Modern radiology fits largely within the first category. Clinical laboratory medicine and anatomic pathology have historically occupied the second. The third is a broad residual category for diagnostic services that do not fit the first two.
CMS now proposes to extract certain computational tests from the second category and place them in the third. Under the OPPS proposal, CMS would remove ten existing codes from the CLFS and assign them to New Technology APCs as SaMS. Under the PFS proposal, CMS would remove the same codes from the CLFS and contractor price them. CMS also proposes that future codes describing stand-alone algorithmic analyses generally follow the SaMS pathway.[8,9]
CMS gives several reasons:
The algorithm operates on data generated by a prior laboratory test rather than directly on physical human material.
The analysis is entirely computer based and, in CMS’s view, can be performed by any unregulated entity possessing the software.
Algorithmic analysis of laboratory data should be treated consistently with algorithmic analysis of radiology images.
CLFS crosswalking and gapfilling do not work well for proprietary algorithms whose development costs are difficult to observe.
The CLFS lacks beneficiary cost sharing and budget neutrality, creating payment and program-integrity concerns.
Some of these are legitimate payment questions. They do not establish that a patient-specific computational pathology test lies outside CLIA.
An H&E whole-slide image is not merely a finalized laboratory value. It is a high-resolution digital representation of the patient’s tissue specimen. An algorithm examining morphology, tissue architecture, spatial relationships, or tumor microenvironment features is computationally examining information derived directly from human tissue. Its output may depend on fixation, staining, slide preparation, scanner characteristics, image resolution, software configuration, patient identification, local workflow, algorithm version, and performance drift. These are exactly the variables CAP addresses through laboratory validation, quality management, and inspection.
CMS’s own CLIA program has previously treated clinical review of digital images as work performed under a CLIA certificate. Its 2023 remote-review guidance left the primary CLIA laboratory and laboratory director responsible for digital review, documentation, reporting, and survey findings.[10]
The statutory payment question
Section 1834A makes the CLFS/PAMA methodology the default for a separately payable clinical diagnostic laboratory test. New CDLTs are assigned CLFS payment through crosswalking or gapfilling. CMS can carve some CLIA-regulated services out of separate CLFS payment—for example, hospital laboratory tests bundled into an OPPS payment, anatomic pathology services paid under the PFS, or tests incorporated into inpatient payments.[11]
But there is an important distinction between creating a defined payment exception and declaring that a clinical test is no longer a laboratory test. If CMS wants to establish a special payment pathway for computational pathology, it should identify the relevant authority and preserve applicable CLIA requirements. Dissatisfaction with PAMA ratesetting, cost transparency, coinsurance, or budget neutrality does not itself convert a CLIA test into an unregulated “other diagnostic test.”
Why CAP should care—and where a consultant could engage [THIS IS LOOK AND FEEL AI DRAFT ONLY]
This is not only a reimbursement dispute. It potentially removes an emerging field from two domains in which CAP has invested substantial institutional authority: laboratory coding and CLIA accreditation.
If computational pathology becomes simply an unregulated SaMS transaction:
CAP’s validation and quality-management requirements could become optional or unclear.
Responsibility could become fragmented among the scanning laboratory, algorithm developer, interpreting pathologist, hospital, and billing entity.
CPT may demand CLIA documentation for a code that CMS subsequently declares requires no CLIA entity.
Laboratories may be uncertain which organization owns the test result and which laboratory director is accountable.
Software vendors may lack a workable Medicare supplier-enrollment and billing category.
CMS could create a regulatory gap precisely as computational pathology begins generating increasingly consequential prognostic and treatment-selection results.
For example, the digital pathology industry has a constructive interest in helping CAP address this issue. The objective need not be to reject SaMS as a concept. Rather, engagement could encourage CAP to propose a clinically credible boundary between:
A genuinely secondary calculation performed on finalized data, which may reasonably be treated as SaMS; and
Computational examination of a whole-slide image or other specimen-derived data to create a new clinical result, which should remain subject to CLIA laboratory oversight.
Coordinated CAP comments on both the OPPS and PFS rules;
Preservation of CLIA oversight even if CMS chooses a different payment method;
A clear framework for distributed computational pathology involving multiple CLIA laboratories;
Allocation of responsibility among the slide-producing laboratory, scanning site, algorithm provider, interpreting pathologist, and reporting laboratory;
Validation and revalidation standards across different scanners, stains, sites, and software versions;
A workable pathway from Category III coding to Category I recognition and national payment; and
Alignment among CAP accreditation, CPT coding, FDA regulation, and Medicare supplier enrollment.
This is a natural issue for CAP leadership. CAP has already asserted that clinical AI belongs within the responsibilities of the CLIA laboratory director. CMS has now proposed a payment policy premised on the opposite conclusion. Policy experts can help CAP defend an important quality role while supporting a practical, innovation-friendly framework for the distributed digital pathology systems that are now emerging.
References
College of American Pathologists. How to Use the New Digital Pathology CPT Codes; see also Digital Pathology Codes.
American Medical Association. CPT Appendix S: Taxonomy for Artificial Intelligence in Medical Services and Procedures.
American Medical Association. CPT Codes Offer the Language to Report AI-Enabled Health Services.
College of American Pathologists. Response to the HHS Health Sector AI Request for Information, February 23, 2026.
College of American Pathologists. CAP Laboratory Accreditation Program.
CAP TODAY. Digital Path Practices Reflected in Latest Checklist Changes and page two, February 2026.
College of American Pathologists. Response to FDA on Measuring and Evaluating AI-Enabled Medical Device Performance, December 1, 2025.
Centers for Medicare & Medicaid Services. CY 2027 OPPS Proposed Rule, 91 Fed. Reg. 41734, especially 41918–41927.
Centers for Medicare & Medicaid Services. CY 2027 PFS Proposed Rule, 91 Fed. Reg. 43842, especially 43910–43912.
Centers for Medicare & Medicaid Services. CLIA Post-Public Health Emergency Guidance, May 11, 2023.
Social Security Act §1834A, 42 U.S.C. §1395m-1. Official statutory text.
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.