There must be some mistake. A company studies its cancer blood test in a randomized trial of 140,000 people using a suboptimal surrogate endpoint, fails to meet that endpoint, yet is recommended for approval by a Food and Drug Administration advisory panel. And because of the Medicare Multi-Cancer Early Detection Screening Coverage Act passed earlier this year, the company appears to be on track to have taxpayers foot the bill.
How does this happen?
The company is Grail, named to reflect the promise of its product: a simple blood test to screen for multiple cancers — the so-called holy grail of cancer screening. Grail has made remarkable claims for its test: that it will save tens of thousands of lives and save money.
It has also been sued by its own investors twice: in 2024 and 2026. Furthermore, the company’s friendly relationship with England’s National Health Service has been criticized — a former Prime Minister David Cameron was a paid adviser, and NHS provided Grail with the public imprimatur of a high-profile trial of 140,000 Britons. All for a test that had shown little promise in early studies.
I have been writing about Grail for quite some time. In 2021, David Carr and I raised concerns about the lack of control group in the initial announcement of the trial (it was later clarified that there was a control group) and later that year were joined by David Kent in an investigation demonstrating that the trial was large enough to test the most relevant question for patients: Will it save lives? In 2022, Barnett Kramer and I questioned the confluence of Congress and multicancer screening and went on to criticize the design of the taxpayer-funded Grail/Medicare study the following year.
I agree that the underlying technology is truly remarkable. This test, as well as others like it, can detect tiny fragments of free-floating DNA in the blood. More remarkably, the technology can distinguish a little bit of tumor DNA from the sea of normal DNA in our blood. It may prove quite useful for cancer patients by helping discern who has been completely treated (no tumor DNA) from those who need more treatment (persistent tumor DNA).
But Grail’s interest is in screening. Why? Because screening means a huge market. While there are only so many cancer patients, all of us are candidates for screening.
Screening is an easy sell. It taps into a widely held belief: the sooner, the better. It always finds more cancer patients, which is interpreted as helping people. It always produces more “cancer survivors” — powerful stories of individuals found to have cancer by the test who are currently doing well. And it is always accompanied by longer survival times and thus higher five-year survival rates.
Easy sell, but horribly deceptive. Telling more people they have cancer is not progress; progress is having fewer people die from cancer. Some survive because treatment works, not because the cancer was detected earlier; others survive because they were overdiagnosed (diagnosed with a cancer not destined to cause symptoms or death). And survival statistics always go up because the survival clock is started earlier and because overdiagnosed patients do really well.
Because of this deceptive feedback, the gold standard for evaluating cancer screening has been a randomized trial of screening vs. no screening to determine the effect on cancer death. That is the standard that has been applied to breast, colorectal, lung, and prostate cancer screening. Even Grail investigators agree: Practice-changing level of evidence requires a randomized trial testing cancer death. Other Grail investigators modeled the likely impact of their multicancer test, estimating a 25% reduction in the death rate for all cancers combined.
But the Grail/National Health Service trial did not investigate the effect of screening on cancer death. Instead, it tested the effect on how many patients were found to have late-stage cancer. That’s a surrogate outcome: While effective screening should reduce the number of patients who present with late-stage disease (because aggressive cancers are found earlier) that doesn’t necessarily mean cancer deaths will fall (because aggressive cancers may be no more treatable when found earlier). That’s what happened with ovarian cancer screening: fewer late-stage patients, but no change in death.
Nevertheless, the Grail/National Health Service trial was negative. No reduction in late-stage cancer. The trial provided no evidence that this test will reduce cancer deaths by 25%. Yet the FDA advisory panel recommends approval.
Don’t ask me why. I don’t pretend to know. But I do know that FDA approval would trigger a tsunami of public and private expenditure.
The reason is the Medicare Multi-Cancer Early Detection Screening Coverage Act, signed into law earlier this year. It mandates Medicare coverage of multicancer screening tests following FDA approval — bypassing the usual role of the U.S. Preventive Services Task Force. And private insurers will undoubtedly face enormous pressure to follow suit.
There was widespread, bipartisan support in Congress. Why? The juxtaposition of the words “cancer” and “early detection” has obvious political appeal. Furthermore, Grail spearheaded a massive lobbying effort — even going so far as to suggest that multicancer screening will reduce health disparities and the American Cancer Society echoed the claim.
Seriously? We can screen away health disparities? I don’t think so.
While the advisory panel has recommended that the FDA approve Grail’s test, the FDA itself has yet to render judgment. Hopefully, the agency will question why it should approve an expensive test whose benefit has not been demonstrated — especially given the prospect of a cascade of coverage decisions bound to make health insurance more expensive at a time when fewer and fewer Americans can afford it.
And with the National Cancer Institute preparing a major randomized trial of multicancer screening tests using cancer death as the endpoint, the agency should also consider whether approving Grail’s test will impede that work. (Why volunteer for the trial if multicancer screening is already covered by insurance?)
Almost forgot: I estimate that screening all Americans 50 and older would have a ballpark cost of $60-$100 billion per year. Here’s the math: There are more than 120 million Americans over 50 — the population for whom Grail recommends annual screening. The test will cost somewhere between about $500 (the current Medicare payment for Cologuard, the benchmark specified in the MCED Coverage Act) and $949 (Grail’s current list price).
To the FDA: Don’t approve the test. Test the tests. Let the National Cancer Institute do its work. And don’t make Americans pay even more for health care.
H. Gilbert Welch is a physician and cancer screening researcher at the Center for Surgery and Public Health at Brigham and Women’s Hospital in Boston.
Source: www.statnews.com
