A blood test found the small liver cancers ultrasound missed entirely

Design

cross-sectional, prospective, blinded, multicentre validation study with multiphasic MRI as reference standard

Population

1,268 evaluable adults with cirrhosis at high risk of hepatocellular carcinoma; 46 (3.6%) with HCC, 46% of lesions 2 cm or smaller

Primary outcome

sensitivity and specificity of the cell-free DNA blood test versus ultrasound

Effect

sensitivity 47.8% (95% CI 32.9-63.1) vs 28.3% (16.0-43.5) overall, and 28.6% (11.3-52.2) vs 0% (0.0-16.1) for lesions 2 cm or smaller; specificity 87.6% vs 93.9%

A cross-sectional, prospective, blinded multicentre validation study enrolled 1,268 evaluable adults with cirrhosis, all of whom gave blood for the HelioLiver Dx multianalyte cell-free DNA test, underwent ultrasound, and had multiphasic MRI as the reference standard. Forty-six (3.6%) had hepatocellular carcinoma, and 46% of those lesions were 2 cm or smaller.

For all lesions, sensitivity was 47.8% (95% CI 32.9-63.1) for the blood test against 28.3% (16.0-43.5) for ultrasound. For lesions 2 cm or smaller the gap was categorical: 28.6% (11.3-52.2) against 0% (0.0-16.1) — ultrasound detected none of them. Specificity was 87.6% (85.6-89.4) for the blood test and 93.9% (92.5-95.2) for ultrasound. The prespecified co-primary endpoints of superior sensitivity and non-inferior specificity were met.

The honest framing is that both tests perform poorly and one performs less poorly. Missing half of all hepatocellular carcinomas is not a surveillance test anyone would design; the comparison is against ultrasound, which missed nearly three-quarters and all the small ones. The specificity trade-off is real: 87.6% in a population with 3.6% prevalence means most positives will be false, each generating an MRI. Where the blood test earns its place is adherence — patients who will not attend six-monthly ultrasound might give blood.

• Do not read superior sensitivity as good sensitivity — the test missed half of all cancers
• Model the false-positive burden before adopting: at 3.6% prevalence, 87.6% specificity generates many confirmatory MRIs
• The reference standard was MRI, not histology, so both tests are being judged against imaging
• Ultrasound detecting no lesions 2 cm or smaller is the finding that should unsettle current practice
• Where MRI capacity is limited — most of Indian hepatology — a test that increases MRI demand needs that capacity planned first

Why it matters

It quantifies how much of the surveillance failure in cirrhosis is the test rather than the patient.

Don't overread it

Cross-sectional diagnostic accuracy against an imaging reference standard — no evidence yet that using this test changes stage at diagnosis or survival.

The statistics, in plain English

A sensitivity of 47.8% with an interval from 32.9% to 63.1% is imprecise because only 46 patients had cancer — the whole sensitivity estimate rests on those 46. The 0% for ultrasound in small lesions has an upper bound of 16.1%, so it means 'no lesions found among a small number' rather than 'never detects them'. Specificity is far better estimated because 1,222 participants did not have cancer.

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.