September 5, 2023 | Medicare, the single largest payer for healthcare in the United States, has a hodgepodge of national and state policies capable of bankrupting innovation in molecular diagnostics. It is an inequitable, fiscally perplexing, needlessly verbose, and fraud-prone system but, if there’s a bright spot, it’s the Molecular Diagnostic Services Program (MolDX) responsible for technology assessments and claims disbursement for more than half the country.
That was the key message of consultant Bruce Quinn, M.D., P.hD., MBA, a self-described “MolDX-ologist,” in his keynote presentation at last week’s Next Generation Dx Summit held in Washington, DC. “You can walk from Washington State to the Gulf Coast of Alabama and never step outside of a MolDX state,” he says, referencing a color-coded map highlighting its dominance.
The other partnering intermediaries for the Medicare program are Novitas (the major player in the south-central U.S.) and the National Government Services Medicare Administrative Contractors (NGS MACs) where Cologuard and Foundation Medicine’s CDx test—both governed by national coverage determinations (NCDs)—are consuming a big chunk of the change.
A pathologist by training who started in academics but has been working in the business and consulting world for more than two decades, Quinn knows what he’s talking about. He even spent several years as a medical director for the Center for Medicare & Medicaid Services (CMS) concentrating on agency policy and innovation, which is also the theme of his longstanding blog.
Quinn begins with his take on the Transitional Coverage for Emerging Technologies (TCET) policy pathway for breakthrough devices recently proposed by the CMS. The public comment period just closed.
TCET is a voluntary application for review of new medical technologies by CMS prior to approval by the Food and Drug Administration (FDA) and is effectively a short window of opportunity since the agencies need to peruse all the same data, he notes. The CMS states in the policy that it expects this will “usually result in coverage with evidence development [CED],” its payment paradigm conditioned on data gathering through a clinical trial or registry to determine a product’s effectiveness.
The agency further remarks that it expects few of the breakthrough products coming through the TCET pathway will have enough evidence for initial direct coverage, Quinn continues, and that the process of moving from a review to a NCD will take about six months. The CMS also states that it expects to take no more than four of these cases per year but “not for diagnostics because they are special contractors... which [though not explicitly saying so] has to be MolDx.”
The TCET pathway is replacing the Medicare Coverage of Innovative Technologies (MCIT) pathway that came out at the tail end of the Trump administration, says Quinn, who never thought of the MCIT as the “parade of horribles” some people envisioned for it. A speech given by Former FDA Commissioner Scott Gottlieb indicated MCIT was thoughtfully developed to reduce the risk of innovative products by permitting national Medicare coverage starting on the date of FDA market authorization.
Some sort of policy is clearly needed, says Quinn, referencing a new paper detailing the lengthy processes required to establish Medicare coverage (JAMA Health Forum (DOI: 10.1001/jamahealthforum.2023.2260) as well as his own “coverage horror story” about a prostate surgery product. It received a two-year new technology add-on payment (NTAP) designation, which involves rigorous CMS review, yet it took an additional two years to finally get local coverage determinations (LCDs) due to a series of backups and delays.
The TCET pathway comes with two guides, one for evidence evaluation that covers principles, evaluating individual studies, and consolidated weight of evidence, and another for CED studies, which covers study deign and the importance of good statistics and outcomes, reports Quinn. He says he is skeptical that CED studies will meet the endpoints of good design, based on his prior experience with a CED program for Alzheimer’s research.
In a public meeting held last year, CMS was “pretty silent” about its rationale for the CED mechanism and focused more on the process, he notes. “There have been dozens of studies logged under CED for 20 to 30 years, under different levels of management, and most of them are not known for having accomplished a lot. There’s no book, there’s no paper on the landmark achievements of CED; it’s conspicuous by its absence.”
Moreover, a few of the larger CED studies are piggybacked on FDA-required post-approval studies, continues Quinn. Opinions have also been circulating questioning the legality of CED. The microeconomics around it are certainly disturbing. “The CMS-proposed or -imposed trial costs can be very large.”
In an NCD proposed for a $2,000 next-generation sequencing in 2017, the CED component would have potentially involved “hundreds of thousands of patients for $10,000 to $15,000 each... and it was absolutely unclear if anyone in the [CMS] Coverage Group had done... [the] back-of-the-envelope math” about how much that would cost,” Quinn says.
The math on CED with a $4,000 per-patient trial cost might work when accompanied by a $25,000 drug payment but not a $100 generic test, he says. In terms of net return on investment, payback period, and discount rate, “CED would generally destroy value and that’s not surprising. It’s just I’ve never seen anyone dollarize it before.” The MCIT program, in contrast, would raise return on investment and net present value by accelerating coverage.
He says that a previous Alzheimer’s study using $4,000-per-patient PET scans on 16,000 participants consumed a whopping $64 million. “Can you imagine how long it would take to get a $64 million NIH study?” The National Oncologic PET Registry study was “five or 10 times bigger.”
One extremely useful incentive the CMS took two years ago was a “forward-looking CED” whereby it covered colorectal blood-based screening tests meeting certain performance criteria, he says. Numerous companies have entered the field who are going after that NCD because the coverage was put together in advance, with all the pertinent details, and they had a few years to reach the performance bars.
On the other hand, CMS has created a “bad incentive” in the exploding arena of chemosensitivity testing where cancer cells are grown in vitro so drugs can be tested against them, he continues. “There is a 40-year-old NCD that bans coverage in that field, and it could take several years to get rid of.”
In digital pathology, meanwhile, a test classed as a physician service will be paid for as part of the physician service, but a test classified as a lab service on an outpatient hospital specimen is by policy always bundled, Quinn says. All the new digital pathology codes put forth by the American Medical Association (AMA) in January are “classed as non-payable in the hospital outpatient setting” and automatically bundled if they are on a tissue specimen.”
At the state level, Quinn zooms in on a “huge fiasco” uncovered by the Office of the Inspector General (OIG) involving overpayments in southern states for some supposedly rare genetic tests represented by AMA tier 2 molecular pathology procedure codes. “These were virtually not paid at all in 2017 [$9,553] because they weren’t priced yet and had to go through manual processing.”
But as soon as the tests had codes, payments ratcheted up year after year, “all through labs that had no business billing Medicare, and they would bill all the tier 2 codes, like 16 rare genes at a time, in hundreds of thousands of Medicare patients,” he says. The fraud was occurring in five states (largely Novitas territory)—Texas, Florida, Louisiana, Pennsylvania, and New Jersey—prompting the Department of Justice to launch Operation Double Helix.
Although a press release about the operation was issued in the summer of 2019, “the vast majority of the fraud took place in the last half of 2019 and continued in 2020 and 2021,” when the annual loss totaled nearly $1 billion. The OIG report came out in July 2023 and talks about the improper payments, which are actually twice that high [$2 billion] because labs generally bill only half their payments under [CPT code] 81408.”
The other half were likely billed under the well-recognized code 81162 for BRCA1/2 testing, and thus presumed to be medically necessary in patients with breast cancer, based on an analysis Quinn did several years ago finding significant overpayments in many of these same jurisdictions that prompted a cease-and-desist letter.
Another debacle with Novitas was a billing article about LCDs for oncology biomarkers, a “vastly longer” version of which was issued this year after the previous draft was suspended amid widespread outcries. “It is horrifying to read... extremely verbose and rambling, [and] parts of it literally make no sense at all,” says Quinn. The document contained 80 overlapping sections of instructions that were reportedly reviewed by the CMS Coverage Group. And such poorly written LCDs continue to appear, he adds.
On the contrary, MolDX has produced some “extremely well-written LCDs,” says Quinn, specifically citing one on squamous cancer tests. “You can understand exactly what they are saying and why.”
LCDs are supposed to have three parts—one each for coverage and limitations, evidence review, and evidence analysis, he points out. But “like Keystone Cops... cannot follow these three simple categories,” he says. One NGS MAC managed to compose a “disorderly and confusing” LCD on the topic of bladder biomarkers in just a page and a half. “You can’t tell what the coverage is after you’ve read it.”
These days, Quinn says he likes to think of molecular diagnostics as falling into one of three categories: standard-of-care tests widely included in guidelines and clinical review articles, novel tests for use cases that are well accepted (e.g., minimal residual disease in colon cancer), and novel tests for a new use case where the risks and benefits are unknown.
The MolDx system not only allows for easier and faster coverage decisions but is also fundamentally different than other payer decision systems, he points out, probably because it was developed by pathologists and lab directors. It is neither a B2B or B2C type model because it plays by an entirely different set of rules, “and it’s to your peril if you don’t appreciate that.”
He is a fan of an approach proposed in the book, $100M Offers: How To Make Offers So Good People Feel Stupid Saying No (Alex Hormozi), which holds that “a good customer will have a lot of pain, a lot of need for your product, have plenty of money to pay for your product, be easy to target, and be a growing market.” The ideal product, meanwhile, is an equation where above the line is the ability to explain “the ideal value times the likelihood of achieving it” and below it is “the time delay to get there and the customer amount of work or money involved.”
The goal is to get the value and the likelihood bigger and the time delay and costs smaller, says Quinn. Blue Cross of Montana might have the money and a growing market for a gene-based pancreatic cancer test, he cites as an example, but is virtually impossible to target. How to get a meeting, who to talk to, what to listen for, what the follow-up should be— “that paradigm for what a business customer needs—completely falls apart when it is a payer.”
MolDX also talks a lot about the combination of data and judgement, he adds. That’s also the rule of thumb for making complex decisions, as detailed in the book Decisions Over Decimals: Striking the Balance between Intuition and Information (Frank, Magnone, and Metzer), just out by a trio of authors from Columbia, Microsoft, and Google.
Quinn says he is fond of the “foundational LCDs” used by MolDX to establish principles and standards around an area such as minimal residual disease testing (MRD) or hereditary disease testing that then get applied on a case-by-case basis. The only potential drawback is that “they can get out of sync with the latest thinking as they age,” leaving one to wonder what the current standard is. As a consultant, one of the questions he is most frequently asked is “What are MolDX’s standards for an MRD test?”
To highlight his point, he cites an LCD for donor organ circulating DNA from April of 2021, which is currently undergoing revisions and will be finalized in 2024. The citation pattern peaks in 2017 with nothing referenced after 2020, meaning “if you want to know what their standards are and how they discuss recent papers, what they like, what they don’t like, what they prefer, what they don’t prefer, it won’t be there... and as the foundational LCD idea goes on further, the primary LCD gets farther in the past and that perhaps becomes an accentuated problem.”
The time delay associated with LCDs of different MACs is also significant, in some cases upwards of two years, Quinn adds, which can drag down net present value.
It is no secret among Medicare medical directors that dossiers are a disaster in diagnostics, says Quinn. “Often the same company that can write a perfectly fine elegant paper for a journal will produce a mess of gobbledygook and confusion in trying to describe itself in a dossier. I think the root cause analysis is that the writer does not fully understand in an agile way concepts like clinical validity and clinical utility.”
Quinn stands behind a paper he co-authored almost a decade ago proposing the key questions that a dossier should address to avoid problems at the CMS review stage. “What is the population and indication? What is the output or value of the current test? What is the output or value of the new test? What is the difference... [and] what change in clinical management should occur based on that delta?”
And finally, he adds, what is the outcome of using the new test in real life? Some reference should also be made to the cost of achieving that.
ChatGPT may ultimately find a role here. Quinn says he has become a big user of the large language model-based chatbot since February and finds it to be “extremely good” at summarizing trade journal articles as well as turning “a cloud of thoughts into logic.” His examples include receiving helpful tips for improving a 20-page report written for a client and composing a statement of work in perfect consulting style almost instantaneously—an onerous task he would normally have sweated over for several days.