Check · Diagnostics · Galleri cancer test In review

Does the Galleri blood test screen for 50+ cancers early and save lives?

Claim attributed to GRAIL (maker of Galleri), plus longevity and concierge clinics marketing MCED blood tests. , GRAIL markets Galleri as a screen for a shared cancer signal across 50+ cancer types. The "saves lives" framing is largely a clinic and press extrapolation; GRAIL's own materials are more careful.

Verdict Unproven

Evidence grade C Low certainty

The signal is real and the technology is promising, but no MCED test has shown it reduces cancer death, and the first large RCT missed its primary endpoint. "Catches 50+ cancers" is partly supported; "saves lives" is unproven.

Detects a real signal in weeks; says nothing yet about whether a screened person lives longer, and the first big trial that tried to show benefit missed its main goal.

The theory

What it’s supposed to target

Circulating tumor DNA (cfDNA)
Methylation signatures
Stage-dependent sensitivity
Positive predictive value

Galleri sequences cell-free DNA shed into the blood and reads its methylation pattern, which can both flag that a cancer signal is present and predict its tissue of origin. The biology is real: dying tumor cells do release detectable DNA, and one blood draw that screens for many cancers at once (a multi-cancer early detection, or MCED, test) is a genuinely attractive idea.

The catch is how much tumor DNA early cancers shed, which is very little. So sensitivity is stage-dependent and low early: high for advanced, aggressive tumors but missing most stage I disease, exactly where early detection is meant to help. Because cancer is rare in any given screened person, even high specificity leaves a modest positive predictive value, so many positives trigger workups that find nothing. Above all, no trial has yet shown it lowers mortality; the large randomized trial is still running. A promising signal, an unproven payoff.

Mechanism is theory, not proof. A plausible pathway explains why something might work, not whether it does. The verdict rests on the evidence below, not the elegance of the theory.

The claim

What would have to be true

The test must reliably detect cancers (HOLDS as a binary signal: specificity ~99.5%).

It must catch them early, when treatment is curative (FAILS: stage I sensitivity is 16.8%).

Earlier detection must shift outcomes, not just stage (UNMET: NHS-Galleri missed its stage III-IV endpoint).

Catching more cancers earlier must translate into fewer deaths (UNTESTED: mortality data pending at 3 and 6 years).

The evidence

What the evidence actually shows

A real signal, but weakest where it matters and prone to false alarms

Galleri reads cell-free DNA methylation patterns in blood. In the CCGA validation set (n=4,077), specificity was a strong 99.5%, but overall sensitivity was only 51.5%, and stage I sensitivity was 16.8% (stage II 40.4%), rising to 90.1% only at stage IV. In the prospective PATHFINDER study (n=6,621), a signal was flagged in 92 people but only 35 were real cancers, a positive predictive value of 38%; 62% were false positives, and 30% of those underwent diagnostic procedures despite having no cancer.

The first RCT missed its primary endpoint; mortality is still untested

The NHS-Galleri trial (~142,250 randomized UK adults) is the first and largest RCT of an MCED test. It did not meet its primary endpoint: combined stage III-IV incidence was unchanged (IRR 1.03, 95% CI 0.92-1.14, p=0.63). A 14% stage IV reduction was a secondary finding only. As independent coverage (ASCO Post) quotes the investigators, "it is not yet known whether...earlier detection ultimately improves survival." Cancer-mortality results are pending at 3 and 6 years.

Evidence quality

Studies, graded, and who paid

Detects a genuine multi-cancer cfDNA signal B Moderate certainty

Methylation classifier flags a real signal across many cancer types; specificity is high (~99.5%).

Catches cancers early (at the curable stage) D Very low certainty

Stage I sensitivity only 16.8%, stage II 40.4%; the test is strongest at late, not early, disease.

Saves lives (reduces cancer mortality) D Very low certainty

Untested for mortality; the one screening-effectiveness endpoint tested (stage III-IV reduction) was not met.

Cited studies with type, size, funding/conflicts, and limitations.
#	Study	Type	Size	Funding / COI	Key limitations
1	CCGA substudy 3 (Klein 2021, Ann Oncol)	Case-control diagnostic validation	n=4,077	Industry-funded Funded by GRAIL; GRAIL-affiliated authors and stockholders.	Case-control design overstates real-world performance; stage I sensitivity only 16.8%; cohort predominantly White.
2	PATHFINDER (Schrag 2023, Lancet)	Prospective cohort	n=6,621 analyzable	Industry-funded Sponsor GRAIL contributed to design, analysis, and drafting; multiple authors GRAIL employees/stockholders.	PPV only 38% (62% false positives); 30% of false positives had invasive procedures; ~92% White; no mortality endpoint.
3	NHS-Galleri RCT (2026 ASCO, GRAIL release)	Randomized controlled trial	~142,250 randomized	Industry-funded GRAIL-authored press release; trial run with NHS England. Headline spins a secondary endpoint despite the missed primary.	Primary endpoint (stage III-IV reduction) not met; full peer-reviewed paper pending; mortality data not yet reported.
4	NHS-Galleri independent coverage (ASCO Post 2026)	Independent news analysis	n=142,942 (trial)	Independent Oncology news outlet reporting on the trial, not the sponsor.	Secondary reporting, not the primary manuscript; confirms stage shift is not proven survival benefit.

A stage shift is not proof of saved lives: finding more early cancers can reflect overdiagnosis (indolent tumors that would never have harmed) rather than benefit, and the trial was not powered for mortality.

See who funded the studies behind every Caveat check

Stay neutral

Unproven ≠ disproven

The mortality question has not been adequately tested yet, and the one screening-effectiveness endpoint that has been tested (stage III-IV reduction) was not met. Unproven is not disproven, but it is not a green light either.

The gap

Where claim and evidence diverge

Demonstrating a survival benefit requires very large, multi-year RCTs with cancer-death endpoints, which is exactly why NHS-Galleri exists and why its mortality data are still pending at 3 and 6 years post-screen. Commercial sale as a lab-developed test has run ahead of that proof.

Follow the funding

The money trail

Galleri lists at ~$949 (self-pay ~$799 or less), is generally not covered by insurance (TRICARE is an exception), and is marketed for repeat annual screening, so costs recur. Concierge clinics bundle it into premium packages, adding incentive to promote it regardless of unproven benefit.

GRAIL funded the pivotal CCGA and PATHFINDER studies and contributed to design, analysis, and drafting, with GRAIL-affiliated authors; it also framed the missed-primary-endpoint RCT as a "substantial reduction in stage IV."

Bottom line

The honest read

Galleri detects a real cancer signal and is a genuinely promising technology, but it is weakest at the early-stage cancers screening is meant to catch, most positive results are false alarms, and it is not FDA-approved (GRAIL submitted its final PMA module in early 2026). Whether it saves lives is unproven; the first RCT missed its primary endpoint and mortality data are pending.

Falsifiable

What would change this verdict

A peer-reviewed RCT (NHS-Galleri or successor) showing a statistically significant reduction in cancer-specific or all-cause mortality from MCED screening.

Independent, non-sponsor data showing high early-stage (I-II) sensitivity and a favorable harm-benefit balance from false positives and overdiagnosis.

Receipts

Sources

Common questions