Historic: A standing state-level body of specialists on-call to MAC
Today: A special committee, often of national experts, brought together by ZOOM for a two hour structured evidence discussion with a dozen questions in advance.
####
Historically (2000-2010 or later), MACs were asked to have standing committees of advisors in 15-20 medical specialties. The appointment process, or duration of standing, were not defined. I believe that the LCD manual - called "IOM PIM Chapter 13" - had a list of specialties either in the body of the chapter or in an appendix.
Today, the 2019 IOM PIM Chapter 13 does not list specialties, and I believe CAC panels are often assembled as small groups of experts for a topical meeting
https://www.cms.gov/regulations-and-guidance/guidance/manuals/downloads/pim83c13.pdf
In addition, today the panels are often assembled to discuss a topic PRIOR to the MAC writing (or revising) an LCD on the topic.
Here is the webpage of past CAC meetings at the Noridian website:
https://med.noridianmedicare.com/web/jeb/policies/lcd/cac
While there is no overall list of standing CAC members (if there are any), and there is no simple list of CAC contributors for a specific topic, each meeting generates a transcript so the number of experts and their names/affiliations, are easy to construct from the transcript.
Noridian held a CAC on July 15, 2024, on biomarkers to risk-stratify patients with ductal carcinoma in situ. There is an AGENDA, a BIBLIOGRAPHY, a KEY QUESTIONS, and a TRANSCRIPT.
https://med.noridianmedicare.com/documents/10546/20409183/071524+CAC+Agenda.pdf
https://med.noridianmedicare.com/documents/10546/20409183/071524+CAC+Bibliography.pdf
https://med.noridianmedicare.com/documents/10546/20409183/071524+CAC+Key+Questions.pdf
https://med.noridianmedicare.com/web/jeb/policies/lcd/cac/cac-transcript-071524
The agenda is trivial (Welcome, Discussion, Closing!).
The bibliography has 30 citations, which were likely provided to panelists in advance.
Likely the most pivotal MAC contribution is the Key Questions, of which there were 12.
The meeting generated a 121-page transcript. AI placed names verbalized in the transcript, into a table:
#
The July 15, 2024 MolDX CAC was not a conventional “yes/no” advisory vote; it was a structured expert conversation designed to help MolDX decide what evidence would be needed for Medicare coverage of biomarker tests that risk-stratify DCIS patients after breast-conserving surgery. Gabriel Bien-Willner opened by saying MolDX wanted to write “the best possible policy” using accurate evidence, and that diagnostic coverage would turn on analytical validity, clinical validity, and clinical utility. He emphasized that the hardest questions were not merely whether a test changes management, but whether it changes management to benefit the patient.
The panel itself was intentionally small despite substantial public interest. It included breast surgeons and radiation oncologists from academic and private-practice settings, including Eileen Connolly of Columbia, Kimberly Van Zee and Atif Khan of Memorial Sloan Kettering, John Williams in private practice in Northern Virginia, Brian Czerniecki of Moffitt, and Alastair Thompson of Baylor. The discussion therefore had a useful mix of radiation oncology, surgery, academic trial knowledge, and real-world practice experience.
The central clinical problem was that DCIS is biologically and clinically ambiguous. It is “cancer” in terminology and patient psychology, but its natural history, mortality risk, recurrence risk, and need for radiation vary widely. The panel generally agreed that current DCIS management is imperfect: radiation after lumpectomy reduces ipsilateral breast tumor recurrence, but often without a survival advantage, and many patients may receive radiation for a small absolute benefit. Atif Khan captured the mood bluntly, saying the current process for deciding radiation in DCIS is “very flawed,” while also noting that use of biological signatures to refine risk is a rational extension of oncology practice.
A major theme was the difference between prognosis and prediction of radiation benefit. Several experts accepted that clinicopathologic factors can estimate recurrence risk, but the more valuable clinical question is whether a patient will actually benefit from radiation. Eileen Connolly made this point clearly: one can predict recurrence risk, but that is not the same as predicting benefit from radiation. She argued that a test suggesting a patient will not benefit from radiation can materially help clinicians and patients avoid treatment, because many patients otherwise say, in effect, “even if my risk is low, I still get some benefit, so I want radiation.”
The panel wrestled with what “low risk” means. There was no single agreed threshold, and experts repeatedly emphasized that low risk depends on patient age, life expectancy, comorbidities, tolerance for risk, and personal preferences. John Williams noted that a “low risk” for a 90-year-old woman might be very different from a “low risk” for a 50-year-old. He suggested that roughly 5% to 10% might be low risk for a general population, while other discussion touched on values around 10% at 10 years or 15% at 15 years.
MolDX pressed the panel toward a policy-usable threshold. Bien-Willner asked what degree of recurrence risk, and what degree of radiation-associated risk reduction, should be considered meaningful enough to justify treatment or coverage of a test. He made clear that management change alone was not enough; the policy question was patient benefit. The discussion eventually centered on absolute, not merely relative, benefit. Connolly noted that with ordinary clinicopathologic features, radiation is often described as giving about a 50% relative risk reduction, but the absolute benefit may be small.
That distinction led to one of the meeting’s most important practical conclusions: a small absolute reduction, around 5 percentage points, may not be clinically meaningful for many DCIS patients. The transcript shows MolDX probing whether a low-risk patient with only modest expected absolute benefit from RT should be treated differently, and the later LCD summarizes the CAC as reaching general consensus that 5% or less absolute reduction in IBTR was a reasonable boundary for “no significant clinical impact” from RT.
There was also a strong patient-communication theme. Kimberly Van Zee described reframing recurrence statistics positively: instead of saying risk falls from 10% to 5%, she tells patients they have a 90% chance of no recurrence, or 95% with radiation. Connolly replied that some patients still focus intensely on the 5% difference, particularly in her majority-minority patient population, where trust and fear may shape decision-making. The exchange made clear that risk tools do not operate in a vacuum; they enter emotionally charged conversations where “small” absolute numbers may still matter deeply to patients.
The panel was divided on current biomarker tests. Some experts saw DCISionRT as valuable because it might identify patients with little or no radiation benefit, which is more clinically useful than generic recurrence prediction. Others were more skeptical. Van Zee argued that before adopting a new biosignature, the field needs published evidence on discrimination, calibration, and actual predictive ability. She also distinguished clearly between the better-established Oncotype DX use in invasive breast cancer and the much less settled evidence base for Oncotype DX DCIS.
A recurring concern was whether biomarker tests add enough beyond existing clinicopathologic tools and nomograms. The experts recognized limitations in tools such as VNPI, MSKCC nomograms, RTOG-style criteria, and physician judgment, but they also worried that biomarker tests might partly repackage or imperfectly improve on those same variables. The later LCD summarizes the CAC as identifying a lack of direct comparisons between biomarker tests and existing scoring systems as a major concern.
Finally, the meeting broadened the endpoint question. IBTR mattered, but some experts placed greater value on invasive recurrence and on predicting relative response to radiation. There was also discussion of “residual risk” groups: patients who might remain at high risk despite standard treatment and therefore may be undertreated rather than overtreated. This is important because a DCIS biomarker could have two opposed uses: sparing low-benefit patients from radiation, or identifying higher-risk patients who need more aggressive care.
Overall, the transcript reads as a sophisticated policy workshop. The experts did not reject biomarkers; most accepted that a genuinely predictive test would be clinically valuable. But the meeting also exposed why MolDX would be cautious: DCIS endpoints are delayed and relatively infrequent, treatment patterns have changed over time, patient preferences are powerful, and current tools must prove incremental accuracy and clinical utility over accessible clinicopathologic methods. The practical policy takeaway was not “biomarkers are useless,” but rather: coverage requires evidence that a test can reliably identify patients for whom radiation provides no meaningful clinical benefit, preferably using a clear absolute-benefit threshold and robust comparison to existing risk-stratification methods.
