Science2 publishers2 min readPublished
FDA panel splits 6-4 on whether Grail's multi-cancer blood test works
The panel was unanimous that Grail's Galleri is safe and divided six to four on whether it is effective, on evidence in which 287 of more than 35,000 people screened got a cancer signal and about 60 percent of those were diagnosed.
The Scientist · Science desk
What happened
- An FDA advisory panel voted 10 to 0 on Wednesday that there is reasonable assurance Grail's Galleri blood test, designed to screen for multiple cancers before symptoms appear, is safe.
- The same ten-person panel split 6 to 4 on whether there is reasonable assurance the test is effective.
- On benefit-risk the panel voted 7 to 2 with one abstention that the benefits outweigh the risk for patients meeting the criteria in the proposed indication.
- A study published Tuesday in Nature Medicine covering more than 35,000 participants found cancer signals in 287 people, about 60 percent of whom were later diagnosed with cancer.
- No multi-cancer early detection test has been approved by the FDA, according to the American Cancer Society, though single-cancer screening tools are already on the market.
Compiled by The ScientistSomething wrong?How this is made
Why it matters
- exposure About one person in every 300 screened gets a signal that no diagnosis follows, and that person still gets a workup: imaging, and sometimes a biopsy.
- constraint There is no published pathway for what to do after a positive result, so each health system would write its own, and Pritchard says the research to base one on has not been done.
- decision The agency is not bound by the vote, so the FDA decides whether a 6-4 efficacy margin licenses the first test of its kind.
- precedent An approval would set the evidence bar that every multi-cancer test behind Galleri gets measured against, and that bar would sit at detection.
Per 1,000 people screened in Grail's study, roughly eight got a cancer signal [2]. About five of those were later diagnosed with cancer, and about three were not [4] [3]. The paper reports more than 35,000 participants, so those rates run slightly high; on the round number, the 287 positives break into about 172 diagnoses and about 115 people whose follow-up did not end in one [1].
Colin Pritchard, a professor and cancer researcher at the University of Washington, said the test may be beneficial for people at high risk of cancer, such as those with a family history, and that for people at low risk false positives could result in unnecessary harms such as expensive imaging or biopsies [14]. "More research is warranted on which populations should be offered this type of testing and on guidelines for follow up after a positive test result," Pritchard said [15].
The 6-4 efficacy vote rests on detection data [3] [7]. Detection answers a narrow question: when the test says cancer, how often is cancer there. Whether the cancer was caught early enough to matter is a separate measurement, taken years later, on stage at diagnosis and on deaths. Ajay Goel, a professor and chair of the department of Molecular Diagnostics and Experimental Therapeutics at City of Hope, said the real goal is not just to find more cancers [13]. "It is to find them early enough to change what happens to the patient," he said [12]. "We still need longer term evidence showing that this leads to fewer advanced cancers and ultimately fewer cancer deaths" [11].
The published account does not report how many cancers the test missed. Sensitivity is the other half of screening performance. What the paper does report is localisation: where a cancer was flagged, Galleri named the area of the body it originated in about 91 percent of the time [8]. That number decides whether a positive result sends a patient to one imaging suite or to several.
Grail emphasised in its FDA application that a negative result does not definitively rule out cancer [9]. Goel called the panel vote "an important step forward for the field" and added: "I would still view it as an encouraging step rather than the final word on the clinical value of this type of testing" [20] [10]. If the agency approves, he said, Galleri should be viewed as "an additional screening tool, not a replacement for established cancer screening" [18].
The FDA is not obliged to follow these panels but often does, STAT reported [17]. The efficacy question drew four no votes; the benefit-risk question drew two, with one abstention [5].
What to watch
- Whether the FDA narrows the indication beyond people 50 and over, or licenses the test as Grail proposed it.
- Publication of a follow-up algorithm for a positive Galleri result, which Pritchard says the research does not yet support.
- Any Grail data reporting stage at diagnosis or deaths.