Podcast episode 105 (in German)

The cancer blood test – three in ten, and why both headlines are true

October 13, 2026

On the same day, it was said of the same study that the cancer blood test had missed its goal — and that it clearly reduces late cancer diagnoses. Both headlines are true. This episode shows how that fits together and which number really counts for you in the end.

What it is about

The test looks in the blood for fragments of genetic material from dying cells and reads their chemical markings; from these it recognizes cancer and usually also the organ it comes from. It was tested in the very first randomized study of a test that looks for many cancer types at once, with 142,250 people in the British National Health Service. The prespecified question, whether stage 3 and 4 diagnoses combined become less frequent, is clearly answered no. Only stage 4 became less frequent — because more cancers already turned up at stage 3. The episode explains why that is not yet a gain in lifetime, what the test does well, and where its gap lies: it finds 3 in 10 cancers.

Key points

142,250 people aged 50 to 77 were assigned by lot; for the primary endpoint, stages 3 and 4 combined, the ratio is 1.03, confidence interval 0.92 to 1.14 — no difference.
Stage 4 alone fell by 14 percent, ratio 0.86, but the confidence interval runs from 0.744 to 0.998 and thus ends just below 1.
In the validation study, the test detected only 16.8 percent of cancers at stage 1, but 90.1 percent at stage 4 — sharpest where it is needed least.
Of 1,801 positive results over 3 rounds, cancer was confirmed in 937, that is, in 52 percent; the false-positive rate was 0.45 percent.
The suspected organ of origin was correct in 92.5 percent of hits, including cancer types without any screening such as pancreas and ovary.
Of all cancers diagnosed in the year after a blood draw, the test found 30.7 percent, for the 12 selected types 54.7 percent.
The episode awards human evidence 9, safety 8, use 6 and hype 9 — and for the first time leaves the direction open, because the decisive question was not asked.

Where the data stop

The study has not a single mortality endpoint, neither cancer-specific nor overall. The shift from stage 4 to stage 3 is a surrogate measure, and every form of early detection shifts the beginning of a disease, not necessarily its end. The decline in stage 4 is moreover a secondary finding after a missed primary endpoint and is therefore not protected against chance. At the time of the episode, the study had not been published in peer-reviewed form; all figures come from the manufacturer’s press release, traced by an independent specialist editorial team.

In practice this means: an unremarkable result is no all-clear, and the test does not replace normal screening. Follow-up on mortality and costs is ongoing; the test is not approved anywhere. Forum reports mention a price of 800 dollars without reimbursement and show that the real burden often only begins with the work-up of a positive result — they do not establish a figure. And anyone with symptoms does not wait for a blood test.

Listen (in German)

Listen on Spotify Listen on Apple Podcasts Listen on Amazon Music

Published on October 13, 2026.

The topic in the database

A blood test that looks for many cancer types at once. The first randomized study missed its primary goal.Cancer blood test (multi-cancer early detection)

All episodes in the app (German)

Information only, not medical advice and not a usage or dosage recommendation. Prescription-only and unapproved substances belong in the hands of a physician. Last updated: 2026-09-22.