A multicancer blood test does not cut late-stage cancer

Design

Randomised controlled trial, England (NHS-Galleri)

Population

142,250 adults aged 50–77 without known cancer

Primary outcome

Stage III or IV cancer across 12 types after three annual screening rounds

Effect

Incidence rate ratio 1.03 (95% CI 0.92–1.14), P=0.63; stage IV 0.86 (0.74–1.00)

NHS-Galleri randomised 142,250 people aged 50 to 77 in England to have blood tested with a multicancer early-detection assay or to have it stored, alongside usual care, at up to three annual rounds. The question was whether adding the test catches cancer earlier across the population.

After three rounds it did not. Stage III or IV cancer across 12 prespecified types — the primary endpoint — occurred at essentially the same rate in both groups (incidence rate ratio 1.03, 95% CI 0.92–1.14). A secondary endpoint, stage IV cancer alone, was lower with testing (incidence rate ratio 0.86, 95% CI 0.74–1.00), but the interval just touches 1.0 and this was not the trial's primary measure.

For the clinician fielding questions about these privately marketed tests — increasingly advertised in India at high cost — the message is that they do not replace bowel, breast or cervical screening and have not yet been shown to shift cancer to an earlier, more curable stage.

• A multicancer blood test did not reduce stage III or IV cancer diagnoses over three annual rounds.
• The stage IV signal was borderline and secondary, not the trial's primary result.
• Advise patients that these tests add to, and do not replace, standard screening programmes.
• Longer follow-up is planned before the clinical value is settled.

Why it matters

It tempers the marketing around blood-based cancer screening with a large randomised result most patients will not have seen.

Don't overread it

The lower stage IV rate was a secondary endpoint with a confidence interval touching 1.0 — hopeful, but not a demonstrated benefit.

The statistics, in plain English

An incidence rate ratio of 1.03 with an interval spanning 1.0 means no detectable effect on the primary endpoint; the stage IV figure of 0.86 reaches but does not clear 1.0, so it cannot carry a firm conclusion on its own.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.