FDA Panel Votes 6-4 That Galleri Cancer Blood Test Is Effective, After Its Own Documents Raised Questions

Photo: U.S. Food and Drug Administration (public domain) — The Great Room in Building 31 on the FDA White Oak campus, where the advisory committee meets.

A Food and Drug Administration advisory panel voted Wednesday, Sept. 23, on three non-binding questions about GRAIL’s Galleri blood test for cancer screening in adults 50 and older. The panel found the test safe, 10-0; found it effective, 6-4, the narrowest margin of the three votes; and found its benefits outweigh its risks, 7-2 with one abstention, according to CancerNetwork’s report on the meeting. FDA itself will decide later whether to approve the test, which is already sold for $949 — GRAIL’s own list price — though not yet FDA-approved. The vote came a day after the two studies behind the application, PATHFINDER 2 and the UK’s NHS-Galleri trial, were published; the NHS-Galleri trial did not meet its main goal.

What the panel decided, question by question

The FDA’s Molecular and Clinical Genetics Panel of the Medical Devices Advisory Committee met from 9 a.m. to 6 p.m. ET at FDA’s White Oak campus in Maryland to “discuss, make recommendations, and vote on information regarding the premarket approval application (PMA) for the Galleri® test sponsored by GRAIL Inc.,” per FDA’s meeting notice. It voted on three questions, quoted verbatim from FDA’s Voting Questions document, with the result CancerNetwork reported from the meeting:

“Is there reasonable assurance that Galleri is safe for patients who meet the criteria specified in the proposed indication?” — 10-0 yes.
“Is there reasonable assurance that Galleri is effective for use in patients who meet the criteria specified in the proposed indication?” — 6-4 yes.
“Do the benefits of Galleri outweigh the risks for use in patients who meet the criteria specified in the proposed indication?” — 7-2 yes, 1 abstention.

The vote is a recommendation; FDA “generally follows the recommendations but is not legally bound to do so,” and the agency has not announced when it will decide on the application.

The number in the headlines, and what sits under it

This week’s coverage has centered on one line from the new paper: adding the blood test to USPSTF grade A/B-recommended screening “increased the number of cancers detected by screening approximately 6.5-fold,” the paper states — and, in the same sentence, by approximately three-fold when added to grade A/B/C screening combined. Behind that figure: during the 12 months after enrollment, 440 participants were diagnosed with cancer — 173 found by the test, 31 by USPSTF grade A/B screening, which is recommended for particular groups within the age range based on factors such as sex and personal risk rather than for everyone 50 and older, and 60 by grade C screening, left to individual choice, like prostate-cancer screening — and 176 (40%) were clinically detected — found outside screening.

The paper also reports three performance figures: positive predictive value (PPV) — the share of positive results that were real cancers — was 60.3%; specificity — how often the test correctly came back negative in people who did not have cancer, the mirror image of the false-positive rate — was 99.64%; and episode sensitivity (the share of cancers found within 12 months that the test had flagged) was 39.3%.

GRAIL’s proposed product labeling, quoted in FDA’s documents, states: “A test result of No Cancer Signal Detected does not rule out the presence of cancer, and individuals should continue with guideline-recommended single-cancer screening tests.”

The paper also makes the case for the test: 126 of the 173 cancers it found — 72.8% — were types with no USPSTF grade A or B screening recommendation at all.

Among the 35,335 participants followed for safety, 213 — 0.6% — had one or more invasive procedures after a positive result; across those participants, 295 invasive procedures took place in total, and 267 of them — 90.5% — were nonsurgical.

Where FDA’s numbers differ, and why

FDA’s tables, in its Executive Summary for today’s panel, and the paper published Tuesday both describe PATHFINDER 2 — the same study. FDA’s tables cover the version GRAIL sells now, against a different analysis set; the paper says it evaluated “an earlier version of the MCED test now replaced by a commercially available version.” Neither document corrects the other.

In FDA’s tables, PATHFINDER 2 has a 12-month episode sensitivity of 35.0% and a PPV of 77.0%; for the NHS-Galleri trial, a separate UK study of the same test, FDA’s own bridged analysis lists sensitivity of 31.6% and PPV of 66.2%. The NHS-Galleri trial’s own published paper reports different figures for itself: episode sensitivity ranging from 26.7% to 37.2%, and PPV of 58.0%, 50.4% and 45.8% across its three annual screening rounds.

FDA’s tables also break results down by cancer stage. Stage I is cancer found while still confined and most treatable; stage IV has spread to distant parts of the body. The test flagged 21.4% of stage I cancers in the US study and 13.6% in the UK study, rising to 63.2% and 60.0% by stage IV.

FDA’s Executive Summary adds its own caution about those sensitivity figures, verbatim: “True cancer status, including disease stage, at the time of the Galleri test is unknown for cancer patients not detected by the Galleri test, and there is uncertainty in the extent to which 12-month episode sensitivity is representative of true test sensitivity.” In plain terms: a cancer the test missed is staged only when it surfaces some other way, which can be later than the blood draw.

The question FDA put in writing

FDA’s Executive Summary also asks the panel what these numbers mean for the product’s name. Discussion question 1(b), quoted verbatim: “Please discuss whether Galleri performance by stage (overall and per cancer type) supports the proposed indications for use of ‘early’ detection of multiple types of cancer. i. If no, please discuss whether Galleri performance supports an alternate indication for the detection of multiple types of cancer, absent the ‘early’ characterization.”

The randomized trial published the same day

The NHS-Galleri trial randomized 142,250 people in the United Kingdom, ages 50 to 77, to receive the blood test or not. It set out to show fewer cancers found late — stages III and IV — than in the group not offered the test; it did not meet that prespecified primary endpoint. Its results paper states, in its own abstract: “We reported elsewhere that the primary endpoint of a reduction in the incidence of stage III/IV cancer diagnoses in the intervention arm versus control arm was not met.” PPV fell across the trial’s three annual screening rounds: 58.0%, then 50.4%, then 45.8%.

OHSU described the result this way: “While the trial did not find a statistically significant reduction in the combined number of stage 3 and stage 4 cancers, a favorable trend emerged over time.” According to OHSU’s release, Nabavizadeh — the lead author of PATHFINDER 2, who was not involved in the UK trial — said the combined stage 3 and 4 result likely reflects a feature of any first year of population screening: a reservoir of previously undetected, asymptomatic stage 3 cancers that the test surfaces all at once.

Who paid, and what a reader can do now

GRAIL designed and funded the PATHFINDER 2 study; OHSU’s release states the funder “also participated in data curation, data analysis and interpretation, and drafting and approval of the publication.” The same release discloses that Nabavizadeh “has served in a consulting/advisory role for GRAIL, Inc. and Exact Sciences and has received honoraria from Roche Diagnostics and MJH Life Sciences,” and states he has no stock or equity interests related to this work.

The Galleri test is already for sale. GRAIL’s cost page lists $949, a self-pay price as low as $799, and a temporary $150 discount through Oct. 2. Medicare does not cover it, nor do most commercial insurance plans. No US guideline body recommends this kind of blood test for screening; the American Cancer Society says accuracy remains uncertain, and USPSTF has no recommendation on them.

What happens next

The panel’s votes are a recommendation, not a decision; FDA has not announced a timeline for ruling on GRAIL’s application. Individual panelist comments and any FDA or GRAIL statements from the meeting had not been reported as of this update — check back for panel discussion detail as more outlets publish their own accounts of the proceedings.