Biohack last covered Grail’s Galleri test in September, when an FDA advisory committee voted on it (read that report). What is new now is a published study from a different angle: what happened when the blood test was added to normal cancer screening in tens of thousands of older adults.
In the PATHFINDER 2 study, adding the Galleri blood test to routine screening increased the number of cancers found through screening 6.5-fold, according to results published in the journal Nature Medicine1. The study enrolled 35,878 adults aged 50 and older in the United States and Canada, and every one of them got the test1. Because nobody was randomly assigned to go without the test, the results do not show that Galleri reduces advanced cancers or deaths1.
What the study found
Galleri is meant to screen for more than 50 cancer types from one blood sample1. It uses a machine-learning program to look at chemical patterns in small pieces of DNA floating in the blood, trying to tell DNA from cancer cells apart from DNA released by healthy cells1. If it finds a cancer signal, it also guesses where in the body the cancer may be, to guide follow-up tests2. A positive result means a signal was found, not that cancer has been diagnosed1.
The study tested Galleri on top of the usual screening advice for four cancers: breast, colorectal, cervical and lung1. Because everyone got the blood test, the researchers compared the cancers found with both approaches against the number expected from routine screening alone1.
Of the 35,878 people enrolled, 32,007 were included in the main analysis of how well the test performed2. The key numbers:
- Positive results: 287 people got a cancer signal2.
- Confirmed cancers: 173 of those 287 were diagnosed with cancer, or 60.3%2. That means roughly 4 in 10 positive results did not lead to a cancer diagnosis.
- False alarms in people without cancer: the test’s specificity was 99.64%, which corresponds to a false-positive rate of 0.36%1.
- Location guess: when the test was right about cancer, one of its two top guesses for where the cancer started was correct in 91.3% of cases2.
- Range: the test picked up cancer signals across 17 broad cancer types, including ovarian, lung, breast and colorectal cancers2.
How early, and which cancers
Finding cancer earlier is the main hope for a test like this, because early cancers are more often curable1. Stage information was available for 151 people with at least one new cancer found by the test2. Of these, 53% were stage 1 or 2, and 70.9% were stage 1, 2 or 32. A university release described the same results as “nearly half” at stages 1 or 2 and “nearly three-quarters” at stages 1 through 31.
Most of the extra cancers were types that have no routine screening. Reports put that share at 70%1 or, more precisely, 72.8% of cancers found by the test in cancer types without a grade A or B screening recommendation from the U.S. Preventive Services Task Force, the panel that sets U.S. screening advice2.
Safety follow-up covered more than 35,000 people1. After a positive result, 0.6% had an invasive procedure, and most of those were not surgery1. This matters because the test is meant for people with no symptoms, and every positive result starts more testing1.
The release also described two patients from Oregon. Franklin Weingarten, a retired physician who joined at 84, got a liver-area signal that led to a diagnosis of a rare bile duct cancer; surgeons removed part of his liver, and at 88 he was reported cancer-free1. Bart Dawkins, 68, was diagnosed with an early liver cancer after a signal; he later developed a separate cancer elsewhere in his liver and was being evaluated for a transplant1. These are individual stories, not evidence about how the test performs overall.
What it does not show yet
The study had one group and no comparison group, so it cannot directly compare the blood test with standard screening or with no screening2. Its measures were descriptive, without formal statistical tests of a set hypothesis, so the numbers describe how the test performed rather than proving a benefit from screening2.
That leaves the biggest questions open. Finding more cancers, and finding many of them early, does not by itself show that fewer people end up with late-stage cancer or die of it1. The study design cannot say what would have happened to the extra cancers without the test1. The researchers said randomized trials, where people are assigned by chance to be tested or not, and longer follow-up are needed to learn whether the test is useful in routine care2.
There is also a version issue. The study evaluated an earlier version of the test that has since been replaced by the version now sold2. And the investigators stressed that Galleri is meant to add to standard screening, not replace it1.
Who is behind it
Grail funded and designed PATHFINDER 21. The company also took part in collecting and analyzing the data, interpreting it, and drafting and approving the publication1.
Oregon Health & Science University’s Knight Cancer Institute reported the results; its team was the study’s largest enrollment site, with about 6,100 participants1. Mayo Clinic was another participating center, and its medical oncologist Karthik Giridhar was an investigator and senior author2. Giridhar disclosed consulting and advisory ties with Grail and other organizations, and travel expenses from Grail2.
How it fits the wider debate
Blood tests that look for many cancers at once are drawing close public scrutiny. On the same day the coverage above appeared, Jenn Dowd, PhD, and Gillian Goddard posted a recorded video discussion on Dowd’s Data for Health newsletter about the data behind new cancer screening tests3. Dowd had earlier written a post on Galleri titled “A New Blood Test to Screen for Cancer: Hope or Hype?”3.
The new numbers point in two directions at once. The test’s low false-alarm rate in people without cancer and its reach into cancers with no screening option stand out2. But a positive result did not mean cancer for everyone who got one2, and each positive result leads to more tests for people who felt healthy1.
What to watch next
- Randomized trials: the researchers said these, with longer follow-up, are needed to show whether testing changes outcomes2. Results showing fewer late-stage cancers or deaths would answer the question this study could not.
- The current version: how the newer, commercially sold version of the test performs, since this study used an earlier one2.
- Regulators: how these published results figure into the FDA’s review, which Biohack covered in its September report.
