- Design
- Retrospective cohort with training and validation sets, nomogram and survival analysis
- Population
- 315 patients with surgically confirmed small hepatocellular carcinoma (3 cm or less)
- Primary outcome
- Prediction of macrotrabecular-massive subtype and 3-year recurrence
- Effect
- Model AUC 0.73-0.75; two or three features linked to recurrence (hazard ratio 2.55, 95% CI 1.24-5.22)
Small hepatocellular carcinomas (3 cm or less) are usually treated with good expectation of cure, but the macrotrabecular-massive (MTM) subtype recurs more and is hard to identify before surgery. This study asked whether Gd-EOB-DTPA-enhanced MRI can flag it noninvasively.
In 315 patients with resected small HCC, three MRI features were independent predictors of the MTM subtype: a hypovascular component of at least 20%, an intratumoral artery, and an incomplete capsule. A model combining them reached an area under the curve of about 0.73 to 0.75, with validation-set sensitivity 77.4% and specificity 72.4%. Having two or three of these features was associated with higher three-year recurrence (hazard ratio 2.55, 95% CI 1.24 to 5.22) and significantly lower recurrence-free survival.
For practice, these are features already visible on a standard preoperative liver MRI. Reporting them flags a small HCC likely to behave aggressively, which can inform how closely the patient is followed after resection and the threshold for considering additional treatment.
- In 315 patients with small HCC, three Gd-EOB-DTPA MRI features predicted the aggressive MTM subtype.
- The features were a hypovascular component of 20% or more, an intratumoral artery, and an incomplete capsule.
- A combined model reached an AUC of about 0.73 to 0.75 (validation sensitivity 77.4%, specificity 72.4%).
- Two or three features were linked to higher three-year recurrence (hazard ratio 2.55, 95% CI 1.24 to 5.22).
- Report these features on preoperative MRI to flag a small HCC that may warrant closer follow-up.
Why it matters
It lets a routine preoperative MRI identify the small liver cancers most likely to recur, informing follow-up intensity rather than waiting for recurrence to declare itself.
Don't overread it
This was retrospective with moderate accuracy; the features raise or lower suspicion and inform follow-up, they do not by themselves change the operation or prove the subtype.
The statistics, in plain English
An AUC around 0.75 is moderate, so these features shift the probability of aggressive disease rather than confirm it; the recurrence hazard ratio of 2.55 with an interval from 1.24 to 5.22 is significant but wide, reflecting the retrospective sample.
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.

Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for radiology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free