Woah! Are Phase 3s no longer enough for CMS?

March 17, 2023

On February 22, CMS announced that they would not be reconsidering the National Coverage Determination (NCD) that effectively denies Medicare patients access to lecanemab, a drug with accelerated approval for Alzheimer’s disease. CMS’s rationale was that “there is not yet evidence meeting the criteria for reconsideration.”

Many in the drug development industry are probably numb to all the policy news relating to the IRA, NCDs, PBMs, etc. The acronyms are endless and the news seems to always be bad, and yet the sun still rises and biotech stock prices have not dropped to zero so it would seem that maybe it’s the usual news cycle blowing everything out of proportion.

And yet, it’s worth us – all of us as would-be patients and a subset of us biomedical innovators and investors – pausing to consider the highly unusual CMS precedent that is playing out in Alzheimer’s.

What’s a National Coverage Determination?

When the FDA approves a product, whether a drug or a medical device, usually CMS automatically covers it. Patients may not notice much of a delay between FDA approval and their ability to get access. But sometimes, CMS wants to take a closer look at the nature of the approval to decide if it wants to apply any conditions before covering the product.

That can be understandable. For example, some medical devices are approved based on their similarity to other approved devices without randomized controlled studies proving their efficacy. So if CMS reviews those cases and decides that it will only cover the device if a patient is enrolled in a clinical study or registry, that’s reasonable. It’s not like it’s refusing to cover the device; it just wants to know if the device really offers clinical benefit.

When gathering more knowledge is essential, then it must be gathered. But when it’s nice to have but not essential, then we should consider the cost of gathering that knowledge. Requiring that patients be enrolled in a study to access the product can create barriers to patients getting access to it and impacts a product’s commercial success, which in turn feeds back to investors’ interest in funding development of such products.

*RA Capital has developed a variety of short, free, self-guided online learning modules called* ***RA University****. Our goal is to share practical and theoretical concepts fundamental to the biotech industry that aren't being presented anywhere else.*

Sometimes all CMS requires is that a treated patient be enrolled in a registry so that their progress can be tracked. Those may only require a fairly light touch as simple as a retrospective enrollment in a patient registry based on medical claims, as is the requirement for leadless pacemakers that have been used in registries and trials aiming to enroll over 70,000 patients since the NCD was instituted in 2017. Other registries require prospective tracking of clinical data and extensive paper forms. Some academic hospitals we spoke with who are involved in NCDs for cardiovascular devices employ multiple full-time staff whose sole job is to manage data tracking for registries. So the kind of study CMS requires can serve as a higher or lower barrier to access for patients.

Drugs tend to be approved based on extensive clinical data from controlled studies showing how well a drug works and in which patients it works, so drug NCDs have been rare and NCDs with the requirement for further study even rarer. In fact, the opposite seems to be more common.

Most drug NCDs have been issued when there is concern about inconsistent coverage by different health insurance plans for a new treatment, such as UnitedHealthcare’s request for CMS to issue an NCD for CAR-T, which CMS did, and ViiV Healthcare’s request they issue one for Pre-exposure Prophylaxis treatments for HIV, for which CMS recently solicited comments. In such cases, NCDs can promote patient access by ensuring that private Medicare insurance plans cannot impose authorization criteria that deny patients access to beneficial treatments.

But sometimes CMS seems to have doubts about the quality of evidence that led to FDA approval. The best example of a drug NCD that was issued because of questions about quality of evidence was the 2011 NCD for Dendreon’s cancer vaccine Provenge.

In 2010, the FDA approved Provenge, and investors were then surprised to learn that CMS would review the quality of evidence and issue an NCD. It seemed like maybe CMS was second-guessing the FDA, a worrisome prospect for anyone investing in R&D thinking that the FDA was the real gatekeeper to commercial success.

Then, in early 2011, CMS issued its NCD functionally affirming the FDA’s decision and authorizing coverage in the same patient population for which the FDA had approved the drug with no deviation from the label. Investors and innovators who had been holding their breath could now exhale. The episode sent the message that CMS, though it might sometimes do a double take, ultimately is aligned with the FDA’s judgment of what constitutes evidence of clinical meaningfulness.

An Alzheimer’s drug class NCD: The beginning of something new

In June 2021, the FDA’s accelerated approval of Biogen’s anti-abeta antibody Aduhelm proved highly controversial. The drug had been tested in two Phase 3 trials that, when pooled together, were stopped for futility (which means that statisticians saw little sign that continuing the trial would lead to a positive result). Later, when data for the trials were analyzed separately, one trial showed that the drug was superior to placebo and the other trial showed little difference. The FDA saw enough evidence of efficacy that it granted accelerated approval and required that Biogen run a confirmatory study.

Critics assumed that millions of patients would want to be treated with an unproven expensive drug that would bankrupt Medicare. CMS announced that it would review Aduhelm and issue an NCD. The NCD that CMS published in April 2022 was unlike anything they had ever put out before.

To start with, CMS said that Adulhelm would only be covered for patients enrolled in randomized controlled clinical studies approved by CMS. Were that all, it would have at least aligned with the medical device precedent, even though those trials tend not to require blinding and randomization to a placebo arm.

But the NCD then went on to declare that going forward anti-abeta monoclonal antibodies that received accelerated approval would only be covered if patients were enrolled in a clinical study and that any antibody approved based on outcomes data would be covered if patients were enrolled in a clinical trial or maybe even just in a registry. This is called “Coverage with Evidence Development,” or CED.

This was disturbing for two related reasons. The first was that the NCD applied to drugs that CMS could know nothing about. This created uncertainty not just for Adulhelm, which at least can’t be uninvented, but for all future drugs in the anti-abeta class that had yet to come to market and depended on investors to fund them. The second is that the NCD suggested that CMS might impose CED requirements (i.e., additional clinical trials) even for drugs that the FDA fully approved based on clinical studies demonstrating outcomes improvement. In other words, CMS was declaring that it was the new gatekeeper and arbiter of which Alzheimer’s drugs would be covered and based on what data, but investors and innovators would not know what those requirements were until after funding the drug through FDA approval.

Anyone who doesn’t understand how investors and companies decide which expensive and risky R&D projects to pursue might be forgiven for not appreciating how groundbreaking and problematic this precedent is.

It’s as if the World Series were decided not by the final scores but by a panel of cricket fans six months after the final game.

Over many decades, the FDA has taught industry its requirements for approving a drug through considerable written guidance and thousands of case studies. Even still, investors must contend with the risk that the FDA’s requirements might change from the time they start funding an R&D program to when the drug actually comes before the FDA for review a decade later. Along the way, there are many FDA interactions that guide companies and investors in their decisions about whether to keep funding a drug’s development.

Yet here was CMS saying that it reserved the right to impose nebulous and potentially onerous requirements of its own, with essentially no case history to guide companies and investors as to how CMS would make its decisions. The process becomes unpredictable. Investors shun uncertainty and here was the greatest and most expensive uncertainty, coming after all R&D investments have been risked. If CMS’s goal was to discourage companies and investors from funding further development of anti-abeta antibodies, it made itself pretty clear.

But we and others hoped that CMS might be persuaded to rescind its Alzheimer’s NCD once it had more data on the anti-abeta class. At around the same time that CMS issued its Aduhelm NCD, other anti-abeta antibodies were undergoing rigorous, large, well-controlled, double-blinded randomized studies that were due to read out shortly. If those failed, everyone would assume that Adulhelm also probably hadn’t really worked. If those other trials were positive, then CMS would have its data and might recognize that the anti-abeta class really was legitimate and retire its NCD.

CMS doubles down

Besides Aduhelm, Biogen was also working in partnership with Eisai on a second anti-abeta antibody called lecanemab that has somewhat different properties. Since CMS finalized the NCD for beta amyloid drugs in April 2022, Eisai and Biogen, lecanemab’s sponsors, have presented phase 3 results showing a statistically significant slowing of decline in patient cognition and function, as measured by CDR-SB, and published their results in NEJM. On January 6, 2023, the FDA granted accelerated approval for lecanemab.

While the accelerated approval was based on a surrogate marker (full approval will come as early as July 2023 based on the FDA’s review of the published full Phase 3 outcomes data), now CMS also had in their hands published results from a robust Phase 3 outcomes study that shows that lecanemab reduced the rate of Alzheimer’s progression by about a quarter and preserved over a third of patients’ ability to function over an 18-month period.

In their original NCD, CMS had said that they were “committed to quickly reconsidering this NCD once an anti-amyloid [aka abeta] drug has answered the CED questions with quality evidence.” If ever there were a basis for CMS to reconsider its NCD on the anti-abeta antibody class, this was it. The Alzheimer’s Association (ALZ) submitted a request to CMS to reconsider, and CMS flatly declined. Specifically, CMS said in their response to ALZ: "After careful review of the request and supporting documentation, we are making this decision because, as of the date of this letter, there is not yet evidence meeting the criteria for reconsideration."

So it’s not saying that it won’t ever reconsider the NCD. It just wants more data.

But how much data will be enough? Let’s back up to what CMS stated their criteria were for reconsideration. CMS said they were looking for answers to the following CED questions:

Does the antiamyloid mAb meaningfully improve health outcomes (i.e., slow the decline of cognition and function) for patients in broad community practice?
Do benefits, and harms such as brain hemorrhage and edema, associated with use of the antiamyloid mAb, depend on characteristics of patients, treating clinicians, and settings?
How do the benefits and harms change over time?

The lecanemab clinical trial data answer these questions (see slide 21 from this presentation), and so one would think it would merit CMS at least agreeing to reconsider its NCD. The Veterans’ Health Administration determined these data were sufficient to cover their patients two months after lecanemab’s approval. But CMS, which is accountable for the Medicare market that covers the vast majority of affected patients, said it wouldn’t reconsider its position. This is unnerving.

The FDA is expected to decide on whether to grant lecanemab full approval on July 6th, 2023. This review will take into account the full set of phase 3 data as well as data from the Open Label Extension that it was not able to consider for accelerated approval.

It’s possible that this full data set (beyond the published results) is exactly what CMS wants to see before retiring its NCD and returning to the old norms of drug-by-drug selective NCDs. But what if it doesn’t? And what if it demands additional clinical studies? What would it hope to learn?

What if? - Leca in a parallel universe

What if instead of Aduhelm, lecanemab had been the first abeta antibody to receive accelerated approval based on the robust clinical trial data we have now. Would CMS have written the same NCD?

If yes, then it’s understandable why it is electing to leave it in place now. But if no, then why isn’t it reconsidering the NCD based on lecanemab’s data?

It feels like Aduhelm awoke a new power in CMS that it now refuses to surrender. And yet, there’s still a chance that CMS will relinquish that power if it recognizes the danger it poses to innovation.

Until we see what CMS does after lecanemab’s full approval, hope remains that CMS won’t permanently install itself as an untested gatekeeper of biomedical progress. Because if it does, that will herald a reset of biomedical innovation for diseases of aging.

Prepare for the worst, hope for the best: What will CMS do next?

There is a high probability that the FDA will grant lecanemab full approval in July 2023.

CMS will then be asked to do two things:

Decide whether, under the current NCD, it would cover patients if they enroll in
a) A full-blown randomized clinical study, OR
b) A burdensome registry that requires prospective tracking of clinical efficacy measures not captured in claims data, OR
c) A broadly accessible registry based on retrospective claims data that doesn’t stand in the way of widespread adoption and commercial success.
Reconsider this unprecedented NCD, which will lead CMS to either:
a) Refuse to reconsider or else agree to reconsider but affirm the NCD and require patients to be enrolled in a trial or registry, OR
b) Agree to reconsider and retire the NCD, covering the drug like it does most medicines.

If CMS again refuses to reconsider the NCD or otherwise reaffirms it, the biomedical innovation industry and its investors need to recognize that we will have crossed into a new world. The IRA already creates warped incentives to develop drugs focused on younger patient populations. Investors and drug developers do not need any more reasons to shift R&D away from diseases that affect the elderly (and other Medicare populations, like ALS patients).

Considering how many in industry still don’t understand the impact of the IRA on innovation, odds are that we will also collectively fail to recognize the impact of CMS affirming that it’s the new arbiter of clinical meaningfulness. And yet some of us will be paying attention. Eventually others will catch on when they realize that investors are obliged to invest elsewhere.

It’s possible that CMS will refuse to retire or reconsider the NCD but then only require that patients enroll in a broadly accessible registry (like retrospective claims) to get lecanemab treatments covered. That likely wouldn’t hinder lecanemab’s widespread adoption and commercial success. And yet, it will still leave the specter intact of a CMS that, in the future, might stand in the way of access by demanding patients enroll in an onerous trial.

The current NCD already functionally negates the benefits of accelerated approval of anti-abeta antibodies because it has a hard, unyielding CED requirement that patients enroll in a randomized trial. Trials are expensive and labor intensive. Only some clinics have experience conducting trials. Patients in rural settings or who get their care at community clinics will be deterred from making monthly trips to clinical trial centers for their infusions. We find it hard to believe that the restriction will enable CMS to start collecting data from a diverse set of patients, one of their stated goals in the NCD.

Even though lecanemab was approved on the basis of a randomized trial, by leaving the NCD in place, CMS has tied its own hands with the CED requirement that patients be enrolled in another trial. It won’t be able to step down that requirement to a mere registry until full approval.

If CMS’ true intent is to collect real-world evidence on lecanemab, like how long the drug prolongs the need for patients to enter long-term care, they should make it relatively straightforward by enabling retrospective enrollment in a patient registry based on medical claims. This would minimize physician and patient burden and enable access for patients treated in rural and community settings. This approach is already used to monitor outcomes for the leadless pacemaker, which also has an NCD. CMS should also reconsider the NCD that allows only one PET scan per patient and ensure appropriate coverage of the MRIs needed to monitor lecanemab’s safety.

Budget vs value

CMS is undoubtedly concerned about the impact of lecanemab on its strained budget. CMS may even believe that lecanemab is overpriced and merits being restricted via every mechanism within CMS’s power.

And yet, we must hope that some people within CMS recognize that ICER is no arbiter of value. The conventional cost-effectiveness analysis underpinning their methodology underestimates the value of medicines and, in other countries, has served as justification for payors to deny patients access to life-saving treatments. Consider, for example, that conventional CEA ignores that a medicine benefits not only patients but also their caregivers. It also ignores that a medicine will eventually go generic/biosimilar (or, as with lecanemab, have its price reduced after 13 years via the IRA’s Medicare negotiation mechanism) and yet still benefit patients and their caregivers, saving Medicare and Medicaid money on nursing homes and preserving the productivity of caregivers.

Hospitals and nursing homes don’t go generic, but medicines that keep people out of hospitals and nursing homes do. Yet conventional CEA ignores all that.

Another element of value that conventional CEA ignores is the benefit that everyone derives from having their risk reduced. We all fear Alzheimer’s, and yet lecanemab and hopefully more advances to come give us hope that maybe Alzheimer’s won’t be as bad a diagnosis in the future. Consider the peace of mind we have when we know there are antibiotics that will keep a skinned knee from turning into sepsis; those antibiotics allow us to enjoy more of life. Consider how the world retreated from tens of trillions of dollars of economic activity during the COVID-19 pandemic and then restored all that activity once vaccines and antivirals reassured us that the threat was now reduced. Peace of mind from having our risk reduced is valuable.¹

By ignoring many or all the petals of the "value flower", conventional CEA underestimates the value of a medicine to society and functionally tries to talk us into spending more money on increasingly expensive nursing homes for Alzheimer’s patients than on medicines that are only temporarily expensive.

We should definitely worry if CMS assumes the mantle from the FDA as the ultimate arbiter of clinical meaningfulness. We will even be able to quantify the cost of that worry, even if conventional CEA can’t do it. Let’s hope CMS doesn’t take us into that alternate universe. By covering lecanemab under the least onerous registry possible upon full approval, such as using retrospective claims, and acting quickly to reconsider the NCD, CMS will preserve investor interest in funding the next generation of Alzheimer’s drugs, which aim to be even more effective than lecanemab. Thirteen years of branded pricing for lecanemab prior to CMS price setting or biosimilar entry is well worth our collective investment as taxpayers to improve our health, ease our worry, and spare our budgets in the long run.

¹ To better appreciate how relevant peace of mind is to valuing a medicine, consider how the US spends $300-400M/year on smallpox vaccines and antivirals and yet treats no one for smallpox. It’s the purest case of paying for risk reduction. ICER’s math breaks entirely on that one. According to ICER’s formulas, if the US spent even $1 without a single fraction of a QALY to show for its expense, it would be overpaying. And yet, it only counts the benefit that comes from actually treating a patient. It has no notion of risk reduction; people worry about the risk of a smallpox outbreak or pandemic and feel reassured knowing that we have countermeasures, but ICER’s math cannot compute worry. ICER would have the US government stop paying for smallpox countermeasures because it can’t value risk reduction and similarly undervalues lecanemab and concludes that it’s not worth paying for at Eisai’s stated price.

Please click here for important RA Capital disclosures.