Ultrasound-derived fat fraction held up against biopsy across aetiology and habitus

Design

Prospective multicentre diagnostic accuracy study with liver biopsy as the reference standard

Population

255 patients with chronic liver disease of mixed aetiology, mean age 51.3 years, 148 men; 57.25% with no steatosis

Primary outcome

Area under the ROC curve for detecting hepatic steatosis of grade S1 or above, S2 or above, and S3

Effect

AUC 0.914 (95% CI 0.873 to 0.946), 0.950 (0.916 to 0.974) and 0.916 (0.875 to 0.947); optimal cutoffs 7%, 11% and 18%

Two hundred and fifty-five patients with chronic liver disease were enrolled prospectively across multiple centres between October 2024 and June 2025, each with a liver biopsy as reference; mean age 51.3 years, 148 men, and 146 (57.25%) with no steatosis at all. Ultrasound-derived fat fraction, measured with a dedicated probe, gave an area under the curve of 0.914 (95% CI 0.873 to 0.946) for detecting steatosis of grade S1 or above, 0.950 (0.916 to 0.974) for S2 or above, and 0.916 (0.875 to 0.947) for S3. The optimal cutoffs were 7%, 11% and 18%.

The subgroup results are the reason this is worth acting on. Steatosis quantification by ultrasound has historically been undermined by the things that come with real livers — inflammation, fibrosis, and a thick body wall. Here the area under the curve stayed above 0.9 for detecting any steatosis in patients with necroinflammation, in those with advanced fibrosis, in those with a skin-to-capsule distance of 1.9 cm or more, and in the mixed-aetiology group with isolated metabolic dysfunction-associated steatotic liver disease excluded.

For Indian practice this matters more than the average paper on liver imaging does. MRI proton density fat fraction is the reference non-invasive method and is out of reach for most patients here on cost and scanner time, while ultrasound is the first and often only cross-sectional study available outside the metros. A quantitative fat fraction on a machine already in the room, with published cutoffs, changes what can be offered — for diagnosis and, more usefully, for monitoring a patient over years. Confirm your own equipment reports the same metric before adopting the 7%, 11% and 18% thresholds; vendor implementations are not interchangeable.

• Check which fat quantification metric your machine reports before applying the 7%, 11% and 18% cutoffs.
• Use it for serial monitoring as much as for diagnosis — that is where repeated MRI is least affordable.
• Advanced fibrosis and necroinflammation did not degrade performance in this cohort.
• A skin-to-capsule distance of 1.9 cm or more did not degrade it either, which matters in obese patients.
• Report the numerical fat fraction, not just a subjective grade of echogenicity.

The statistics, in plain English

An area under the curve of 0.95 for moderate or worse steatosis means the test almost always ranks a steatotic liver above a non-steatotic one — good discrimination, but it says nothing about how a single reading should be interpreted. That is what the cutoffs are for, and a cutoff always trades sensitivity against specificity; the 7% threshold for any steatosis will label some normal livers as abnormal. The other limit is the reference standard. Biopsy grades steatosis semi-quantitatively from one small sample of a heterogeneous organ, so some of the apparent error in the ultrasound measurement is error in the comparator. A prospective design with biopsy in every patient, rather than only those with abnormal scans, is what keeps these estimates honest.

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.