What the NHS-Galleri Trial Actually Showed (And What It Means For Your Screening)

If you have been hearing about the Galleri multi-cancer blood test for the last three years and waiting for the real evidence to land, it just did. On May 30, 2026, GRAIL presented full results from the NHS-Galleri trial at ASCO. The trial is a big deal: 142,000 UK adults, three rounds of annual screening, head-to-head with standard NHS screening, full pre-registered endpoints. The result is more textured than either side's talking points - and that texture is what matters for the actual screening decision in front of a patient. This article is for patients trying to decide whether multi-cancer early detection (MCED) testing is worth doing for them. We will lay out what the trial did, what it showed, what it did not show, and how we think about Galleri at Fishtown Medicine now that the data is in.

Why this trial mattered before it ran

For most of the history of cancer screening, the model has been one cancer, one test, one program: mammograms for breast, colonoscopy for colorectal, low-dose CT for lung in high-risk smokers, Pap/HPV for cervical, PSA for prostate (debated). For dozens of cancers - pancreatic, ovarian, esophageal, liver - no screening test exists at all. Patients are diagnosed when symptoms appear, often at Stage III or IV, often after an emergency visit. The five-year survival math drives the urgency. Across all cancers in the SEER database, five-year survival is 93% at Stage I, 86% at Stage II, 64% at Stage III, and 26% at Stage IV. The clinical lever - shifting the stage at which cancer is detected - is one of the highest-impact moves in medicine. The promise of MCED testing was to do that across many cancer types simultaneously with a single blood draw. Galleri, developed by GRAIL, uses targeted methylation analysis of cell-free DNA in blood to detect a "cancer signal" shared across more than 50 cancer types. The chemistry had been validated. What had not been proven was whether deploying the test at population scale would actually shift the stage distribution enough to matter. The NHS-Galleri trial was designed to answer that question.

What the trial actually did

Design: prospective, randomized, controlled trial. Enrolled 142,000 demographically representative adults aged 50-77 across eight regions of England. Half received the Galleri test plus standard NHS screening (breast, bowel, cervical, high-risk lung) once a year for three years. Half received standard NHS screening alone. Follow-up: cancers diagnosed from the first appointment through 18 months after the third appointment were captured and staged. Primary endpoint: a statistically significant reduction in combined Stage III + IV cancers in a pre-specified group of 12 deadly cancers (anus, bladder, colorectal, esophagus, head/neck, liver/bile duct, lung, lymphoma, myeloma/plasma cell neoplasm, ovary, pancreas, stomach) - these represent about two-thirds of cancer deaths in England and the US. Secondary endpoints: Stage IV reduction alone, overall cancer detection rate, route of diagnosis (screening vs. emergency), test performance (PPV, specificity, sensitivity, cancer-signal-of-origin accuracy), safety.

What the trial showed

The primary endpoint was not met

Combined Stage III + IV cancers across all three screening rounds: 3% higher in the Galleri arm vs. control (incidence rate ratio 1.03, p = 0.6324) for the 12 prespecified cancers. Not statistically significant in either direction. This is the headline that ran. It is important. And it is incomplete without the context.

Why the primary endpoint missed: a prevalent-round Stage III spike

The Stage III + IV combined result was dragged by an unusually high Stage III count in the prevalent screening round (the first time someone gets screened). The prevalent round detects cancers that were already present in the population at the time screening started - many of them undiagnosed Stage III cancers that the test pulled forward into clinical visibility. The control arm has not yet had the chance to find these cancers; they will surface with longer follow-up, but the trial's analytic window closed before that happened. The trial's chief investigator, Professor Charles Swanton, framed it this way at ASCO: "The trial has revealed just how much undiagnosed and uninvestigated Stage III cancer is already prevalent in the population before any screening commences. Finding these cancers earlier means we can start treating those patients with the urgency needed and, in many cases, with the opportunity of curative intent." In other words: the apparent Stage III "excess" in the Galleri arm is not the test failing. It is the test exposing a baseline rate of late-stage cancer hiding in the population - cancers that would have surfaced as Stage IV or as emergency presentations later in the control arm. The test pulled them into the diagnostic window earlier.

The Stage IV signal is real and strengthens with each round

The cleaner read is on Stage IV alone, where the screening effect should mechanically be largest if the test works:

Round	Stage IV (12 cancers)	Stage III + IV (12 cancers)
Round 1 (prevalent)	9% lower	19% higher
Round 2 (incident)	22% lower	5% lower
Round 3 (incident)	26% lower	12% lower
All 3 rounds	14% lower (p < 0.05)	3% higher (NS)

A few things to notice:

1. Stage IV reduction strengthens with each round. The incident rounds (rounds 2 and 3) are the ones that most closely approximate the steady-state of an annual screening program. The 22% and 26% reductions in those rounds are the number to anchor on for a patient considering annual screening.
2. Stage III + IV combined also flips negative in the incident rounds. Once you get past the prevalent-round backlog, the trend is in the expected direction.
3. The pooled three-round Stage IV result (14% reduction) was nominally statistically significant. This is the strongest standalone effect.

Specific cancers with the biggest Stage IV reductions

The Stage IV reduction was not uniform. The trial showed the largest effects in cancers where Stage III vs. Stage IV survival differs the most - i.e., cancers where catching them one stage earlier is genuinely curative:

• Esophageal cancer: Stage IV diagnoses reduced by 57.1% in incident rounds. Five-year survival: 24.7% at Stage III, 6.2% at Stage IV.
• Colorectal cancer: Stage IV reduced by 34.4%. Five-year survival: 64.2% at Stage III, 11.0% at Stage IV.

For cancers like these, the move from Stage IV to Stage III is the difference between palliative care and a real shot at cure. That is the clinical case for MCED in one sentence.

Stage I-II detection rose by 16%

Across the 12 pre-specified cancers, the Galleri arm had 16% more Stage I and II cancers diagnosed over the trial period. The increase was concentrated in cancer types typically diagnosed late: ovarian, esophageal, pancreatic, and liver/bile duct. About 68% of cancers detected by Galleri were caught in Stage I, II, or III - the stages where treatment with curative intent is most likely to be possible.

Four times more cancers detected through screening

When Galleri was added to standard NHS screening (breast, bowel, cervical, high-risk lung), the number of cancers detected through any screening route was 4x higher than NHS screening alone. By itself this is a significant claim about the marginal value of MCED on top of an existing program: the standard NHS programs detect about 6% of incident cancers in England (about 14% in the US); Galleri quadruples that yield.

Emergency presentations dropped 25%

In the Galleri arm, cancers diagnosed in an emergency setting were 25% lower than in the control arm. This is one of the strongest indirect signals that the test is shifting cancer detection earlier - emergency presentation typically correlates with advanced disease and worse outcomes.

Test performance held up

The Galleri test had three key performance numbers to defend. All three landed:

Metric	Result	What it means
Specificity	99.55%	The test correctly rules out cancer in ~99.55% of people who don't have it.
False positive rate	0.45%	About 1 in 220 people without cancer gets a positive result.
Positive predictive value (PPV)	52%	About half of people with a positive Galleri result are confirmed to have cancer on workup.
Cancer-signal-of-origin (CSO) accuracy	92.5%	When the test predicts where the signal is coming from, it's right 92.5% of the time.
Episode sensitivity (12 prespecified cancers)	54.7%	The test catches just over half of these cancers diagnosed within 12 months of the blood draw.
Episode sensitivity (all cancer types)	30.7%	Across all cancers, the test catches about 1 in 3 diagnosed within the year.

The PPV is the number that matters most clinically. A 52% PPV means about half of people with a positive Galleri result will be diagnosed with cancer. By contrast, mammography's PPV in screening populations sits around 4-5%. CT lung screening in high-risk smokers is around 1-4%. Galleri's PPV reflects the test's high specificity: false positives are uncommon, so positive results are far more likely to be real. A 100-person framing that puts it in perspective: if 200 people take the Galleri test, about 2 will get a positive result. Of those 2, about 1 will be diagnosed with cancer. The other 1 will go through a diagnostic workup that ultimately rules out cancer - a real psychological and resource cost, but not the avalanche of false alarms some critics predicted.

What it does not show

Three things the trial did NOT prove:

1. It did not show reduced cancer mortality. Mortality endpoints take longer than a 3-year screening trial to accumulate. The trial follow-up will extend over coming years; mortality data will be reported then.
2. The combined Stage III + IV primary endpoint was negative at the statistical level set in the protocol. GRAIL did not get the win they pre-committed to. That matters for regulators and for the most conservative read of the evidence.
3. It did not prove benefit in any single individual. Screening trials measure population-level effects. Whether an individual person benefits from being screened depends on whether they would have developed a screen-detectable cancer in the screening window, whether the test would have caught it, and whether catching it earlier would have changed their outcome.

The Galleri test, like all cancer screening tools, has a label disclaimer: "The Galleri test has not been shown to reduce deaths due to cancer or improve overall survival." That is honest. Mortality data is several years away.

How we think about it at Fishtown Medicine

Multi-cancer early detection testing has been a part of our advanced screening conversation for some time. The published guide to our overall screening philosophy lives at Advanced Cancer Screening (Galleri & Prenuvo). The NHS-Galleri results sharpen, but do not fundamentally change, the way we approach the conversation. Who we think MCED is worth considering for:

• Adults aged 50 and older (the on-label population, matching the NHS-Galleri inclusion criteria).
• Patients with strong family history of cancers that lack good single-cancer screening (pancreatic, ovarian, esophageal, hepatobiliary).
• Patients with elevated baseline cancer risk (germline variants, prior cancer history with remission, heavy smoking history, occupational exposures).
• Patients who have done the math on the trade-offs and want maximally proactive cancer surveillance as part of their longevity plan.

Who we are more cautious with:

• Patients with significant health anxiety where a false positive workup would be more harmful than the marginal early-detection benefit.
• Patients aged under 50 - the trial's evidence base does not extend there, and pre-test probability of cancer is lower (which mathematically lowers PPV).
• Patients during pregnancy, under age 21, or undergoing active cancer treatment (per the test's safety labeling).
• Patients who would not act on a positive result - the test only delivers value if the diagnostic workup happens.

What we tell patients to expect:

1. It is additive, not a replacement. Galleri layers on top of standard screening (mammogram, colonoscopy, Pap/HPV, CT lung if you qualify, prostate evaluation). It does not replace any of those.
2. A positive result is not a diagnosis - it is a signal to investigate. About half of positive results are confirmed; the rest involve diagnostic workup that ultimately rules out cancer. That workup typically involves imaging (PET-CT, MRI, or focused imaging based on the predicted cancer-signal-of-origin) and sometimes biopsy.
3. Annual cadence matters. The Stage IV reduction strengthened from 9% in the prevalent round to 26% in the third round. The longer you do it, the better the screening signal performs.
4. The cancer-signal-of-origin prediction is the workup roadmap. A 92.5% CSO accuracy means the test usually points the diagnostic team in the right direction, which substantially shortens time-to-diagnosis after a positive result.
5. It is not covered by insurance and is not FDA-approved. Out-of-pocket cost is meaningful (typically $750-$950 depending on the source). HSA/FSA eligibility varies.

Key Takeaways

• The NHS-Galleri trial is the largest randomized controlled trial of a multi-cancer early detection test ever conducted (142,000 participants, 3 annual screening rounds).
• The primary endpoint (statistically significant reduction in combined Stage III + IV cancers) was not met, dragged largely by a Stage III spike in the prevalent screening round.
• The secondary endpoints showed real and clinically meaningful effects: 14% Stage IV reduction overall (22% in round 2, 26% in round 3), 16% Stage I-II increase, 4x more cancers detected by screening, 25% fewer emergency presentations.
• Test performance held up: 99.55% specificity, 52% PPV, 92.5% cancer-signal-of-origin accuracy.
• For cancers where Stage III vs. Stage IV survival differs the most, Stage IV reductions were dramatic (esophageal 57.1%, colorectal 34.4% in incident rounds).
• The Galleri test is additive, not a replacement for standard screening. A positive result triggers a diagnostic workup, not a diagnosis.
• Mortality data is several years away; this trial measured stage shift, not deaths prevented.
• Reasonable to consider for adults 50+ with elevated cancer risk who want maximally proactive screening and can absorb the out-of-pocket cost. Not recommended for under-21, pregnancy, or active cancer treatment.

Scientific References

1. Swanton C. NHS-Galleri: Primary Results From a Randomised Controlled Trial to Assess the Clinical Utility of a Multi-Cancer Early Detection (MCED) Test in Population Screening. American Society of Clinical Oncology (ASCO) Annual Meeting; 2026 May 29-June 2.
2. GRAIL, Inc. GRAIL Reports Full Results From NHS-Galleri Trial Demonstrating Substantial Reduction in Stage IV Cancer Diagnoses at 2026 ASCO Annual Meeting. Press release. May 30, 2026.
3. Nabavizadeh N, et al. Safety and Performance of a Multi-Cancer Early Detection (MCED) Test in an Intended-Use Population: Initial Results from the Registrational PATHFINDER 2 Study. ESMO Annual Meeting; October 2025.
4. Klein EA, Richards D, Cohn A, et al. Clinical validation of a targeted methylation-based multi-cancer early detection test using an independent validation set. Ann Oncol. 2021;32(9):1167-77.
5. Schrag D, Beer TM, McDonnell CH, et al. Blood-based tests for multi-cancer early detection (PATHFINDER): a prospective cohort study. Lancet. 2023;402:1251-1260.
6. SEER Research Database, 17 Registries, Nov 2024 Sub (2000-2022). Five-year survival by stage.
7. NHS-Galleri Trial public results page. https://www.nhs-galleri.org/

Medical Disclaimer

This article is for educational purposes only and is not medical advice for any individual. The Galleri test has not been shown to reduce deaths due to cancer or improve overall survival, and it is not FDA-approved. The decision to undergo multi-cancer early detection testing is personal and should be made in consultation with your physician, taking into account your age, family history, baseline cancer risk, tolerance for diagnostic workup of a possible false positive, and personal preferences. The Galleri test is intended for adults aged 50 and older with elevated cancer risk and is not recommended in pregnancy, under age 21, or during active cancer treatment.