Enhanced MRI Tops Ultrasound in Catching Early Liver Cancer

Enhanced, abbreviated MRI (AMRI) outperformed ultrasonography (US) as a screen for early liver cancer in high-risk patients with cirrhosis, according to a single-center, randomized clinical trial.

Overall, AMRI yielded significantly more early-to-advanced stage cancer among 759 patients: the per-patient detection rate in Barcelona-Clinic Liver Cancer (BCLC) stage 0, A, B, or C hepatocellular carcinoma (HCC) was significantly greater in those randomized to screening with hepatobiliary-phase image AMRI (HBP-AMRI) using gadoxetic acid than in those screened with US, at 8.5% versus 3.1% (P=0.002).

That screening performance gap widened in early-stage HCC (BCLC stage 0 or A), where per-person detection rates reached 7.7% in those screened with HBP-AMRI, compared with 2.9% of those screened with US (P=0.003). There was a difference even in detecting very-early BCLC stage 0 cases (6.1% vs 0.8%, P=0.001), reported Sung Won Chung, MD, of the University of Ulsan in Seoul, Korea, at the European Association for the Study of the Liver annual meeting in Barcelona.

“The bottom line is that HBP-AMRI significantly outperformed conventional ultrasound for HCC surveillance in high-risk patients with cirrhosis,” Chung said during his presentation of the AMRIUS study. The results “support a risk-stratified surveillance approach.”

Theoretically, improved HCC screening would suggest more people receiving curative treatment and therefore a possibility of a survival benefit.

HCC guidelines from the American Association for the Study of Liver Diseases (AASLD) currently recommend surveillance with a combination of liver ultrasound and alpha-fetoprotein (AFP) testing. Despite emerging research supporting MRI-based surveillance over US, the AASLD doesn’t recommend routine use of MRI surveillance in at-risk patients.

In his report, Chung said HBP-AMRI reduced per-exam false referrals, particularly in patients who were Child-Pugh A, while maintaining surveillance performance across image quality.

Overall, false referrals occurred in 2.1% of the HBP-AMRI group and 4.4% of the US group (P=0.017). Among those who were Child-Pugh A, the rates were 1.5% with HBP-AMRI and 4.3% with US (P=0.004). Child-Pugh B patients saw no significant differences between the two screening approaches.

Of note, variable operator skill, particularly with US, can affect HCC detection rates.

Chung acknowledged concerns that settings with fewer resources may not match the results of his center’s skilled MRI and US screening teams. In other studies of MRI surveillance in early-stage HCC, however, the detection rate was quite similar, Chung noted.

“There may be some limitations in a primary care setting,” he said, “but tertiary care and other centers that treat HCC patients or have a hepatology clinic will have almost similar results.” The gadoxetic acid agent provides high liver-to-lesion contrast, which will make it “much easier for other people to use this protocol,” Chung added.

The AMRIUS trial enrolled 759 adults age 20 years or older who had liver cirrhosis without a history of HCC. Treated at a single center in South Korea between August 2022 and November 2024, patients were at high risk of HCC, with an estimated annual HCC risk greater than 5%. Exclusion criteria included coexisting cancer with a high recurrence risk, a liver disease-severity Child-Pugh grade of C, or an estimated glomerular filtration rate less than 30 mL/min/1.73 m².

Patients were randomized to two rounds of surveillance during 1 year with either HBP-AMRI with gadoxetic acid as a contrast agent, or with US. The study’s primary endpoint was the detection rate of HCC that is BCLC stage 0 or stage A.

Baseline characteristics were similar between the two treatment groups. Median age in the HBP-AMRI group and the US group was 62 and 61 years, respectively, and 56.7% and 55.6% of patients were men. Hepatitis B virus was the primary cirrhosis etiology, at 56.7% of all patients, followed by alcohol-associated cirrhosis (19.6%), metabolic dysfunction-associated steatotic liver disease (5%), and hepatitis C virus infection (4.7%). Most patients were Child-Pugh A (81%).

Among all patients assessed in the trial, an overall 5.8% had HCC detected.

HBP-AMRI significantly shifted the stage distribution at diagnosis toward earlier disease, Chung noted. Among those randomized to HBP-AMRI, 71.9% of detected HCC was very-early BCLC stage 0, compared with 25% of HCC detected from US screening.

HBP-AMRI also showed an advantage over US in image quality. At the second round of surveillance, 66.2% of HBP-AMRI images had no or minimal image limitation, 27.4% had moderate limitation, and 6.4% had severe limitation. For US, the respective percentages were 37.6%, 40.2%, and 22.2%.

Image quality led to significant differences in false referral rates with US, with steep differences depending on whether an image had no or moderate limitations (3.1%) or severe limitations (8.9%, P=0.037).

On the other hand, with HBP-AMRI, false referrals did not hinge on image quality, staying at 2.1% whether images had no or moderate limitation or severe limitations.