Across 142,250 adults in England, three yearly rounds of GRAIL’s Galleri test did not lower stage III and IV cancers, the trial’s main goal. Stage IV cases were 14% lower, but whether lives are saved will take years to show.
The first large randomised trial of a blood test designed to catch many cancers early has failed its main test, leaving the NHS without the proof it wanted before offering it widely. Adding GRAIL’s Galleri test to usual care did not reduce advanced cancer diagnoses, according to the full results, published last month in the New England Journal of Medicine.1 The UK National Screening Committee, which the government says will consider them, next meets on 26 November.212
The stakes are national. NHS England planned in 2021 to extend the test to a further one million people if the trial succeeded.5 In 2024 it chose to wait for these final results instead, judging the first year’s data not compelling enough for a large pilot.4
Why the goal was missed is disputed. The most advanced cancers, at stage IV, were less common in the tested group, but more stage III cancers cancelled out the gain.137 The explanation offered alongside GRAIL’s results is that some cancers heading for stage IV were caught at stage III instead, when more can be cured.711 Professor Richard Houlston of the Institute of Cancer Research told the Science Media Centre on 30 May that this explanation for the stage III rise was “entirely speculative”.11
The trial, published online on 22 September, enrolled 142,250 adults aged 50 to 77 with no sign of cancer, and gave half of them three annual tests on top of usual care.113 Across 12 prespecified cancers, mostly types with no routine screening, such as pancreas, ovary and stomach, there were 706 stage III or IV cases in the test group and 688 in the control group.713 That is a rate ratio of 1.03, with a 95% interval of 0.92 to 1.14: no measurable difference.1
Stage IV was diagnosed at 145.2 per 100,000 person-years in the test group and 168.5 in the control group, a ratio of 0.86.13 Set against the combined rate, that implies stage III ran about a quarter higher with the test. GRAIL said in February that stage III incidence was higher than expected, and that diagnosis after a positive result was getting quicker as doctors gained experience with the test.7
Professor Nitzan Rosenfeld of Barts Cancer Institute put the missed endpoint down to that rise in stage III detection, especially in the first screening round.11 The stage IV gap then widened with each round, though only the third-round interval excludes no difference.13
One reading of these figures is that the test moves cancers one stage earlier, so stage IV falls first and stage III follows as later rounds catch cancers before they get that far. Another is that the test finds cancers sooner without changing their course, a pitfall Rosenfeld flagged as lead-time bias: more time living with a diagnosis, not more life.11 The trial cannot yet tell the two apart. The screening committee said in May that it remains uncertain whether fewer late-stage diagnoses reliably mean fewer deaths across cancer types.3
Professor Charles Swanton, the lead author, said the trial showed yearly testing can work across a national health service, and that it found more cancers.12 GRAIL calls the stage IV result a substantial reduction.7 Julie Gralow, chief medical officer of the American Society of Clinical Oncology, said the trial missed its endpoint but offered hope for cancers such as ovarian and pancreatic that lack screening.12
Houlston said the researchers had presented the findings more favourably than the overall results warrant, and that nothing yet justifies offering the test to the whole population.11 Rosenfeld called the later stage IV fall an untested possibility, and said cost-effectiveness may be less favourable even if benefits are confirmed.11
The 12 cancers in the main measure cause two-thirds of cancer deaths in England, according to GRAIL, and most have no screening programme at all.12 The test did find more cancer. Over the three rounds, 237 more cancers of all types were diagnosed in the test group than in the control group, and about four times as many were found through screening.14 A quarter fewer were diagnosed after an emergency presentation, which GRAIL links to higher mortality.714
Finding more cancers is not the same as helping patients. Some extra diagnoses may be cancers that would have been found later anyway, and some may never have caused harm; coverage of the conference results noted that neither is yet known.14 Dr Ian Walker of Cancer Research UK said, in comments carried by PA on 30 May, that “we do not know if the test could reduce cancer deaths” in people without symptoms.12
There is also a cost in the clinic. About 1 in 100 people tested positive in each round, and between 46% and 58% of them were then diagnosed with cancer, so roughly half of positive results led to investigations that found none during the trial.9 Houlston warned that testing at population scale would put heavy pressure on already stretched diagnostic services.11 The test is not approved for routine NHS use, and Cancer Research UK warns that a positive result can bring worry and tests that turn out to be unnecessary.8
Money rides on the answer too. GRAIL funded the trial, which The BMJ put at £150m. The journal reported in 2024 that NHS England had agreed to buy millions of tests if the results were favourable, in return for a GRAIL facility in the UK; the contract terms are confidential.10
Follow-up after the third round was a median of 17 months, so there are no data yet on deaths, the outcome that matters most.13 It is unknown whether the stage IV fall will turn into fewer deaths, or how many of the extra cancers found are overdiagnosis rather than earlier diagnosis.
The Department of Health and Social Care said on 30 May that the independent UK National Screening Committee would consider the results, alongside wider evidence on cancer blood tests, before deciding on next steps.12 At its June meeting the committee was still waiting for the published paper and was discussing next steps with NHS England and the National Institute for Health and Care Research. It next meets on 26 November.2 NHS England has not published a response to the journal paper.
The committee set out its test in The BMJ in May. A fall in late-stage disease cannot yet replace a fall in deaths as proof, though a significant reduction in a randomised trial might justify early planning while mortality data mature.3 On this trial’s main measure, there was no such reduction.
The trial team plans to reassess cancer stage in 2027 with a further 12 months of data, and to report on cancer deaths in a few years, along with patient anxiety, the effect on NHS services and value for money.6 In the United States, a Food and Drug Administration advisory panel voted 7 to 2, with one abstention, that the test’s benefits outweigh its risks. The agency’s decision is expected in the coming months.15