Background
Hepatocellular carcinoma (HCC) is the most common primary liver cancer and a major cause of cancer-related death worldwide. People with cirrhosis are at especially high risk, which is why regular surveillance is recommended in this population. For years, ultrasound has been the standard screening test because it is widely available, noninvasive, and relatively inexpensive. However, ultrasound can miss small tumors, especially in patients with nodular cirrhotic livers, obesity, or limited acoustic windows.
At the same time, magnetic resonance imaging (MRI) offers much higher soft-tissue detail and can detect lesions that are difficult to see on ultrasound. The challenge has been feasibility: a full contrast-enhanced MRI is more expensive, takes longer, and is not practical for repeated surveillance in every high-risk patient. To address this, researchers have developed abbreviated MRI protocols, including non-contrast abbreviated MRI (AMRI), which aim to shorten scan time while preserving diagnostic performance.
This interim report from a prospective surveillance trial compared a rapid non-contrast AMRI protocol with ultrasound for HCC surveillance in patients with cirrhosis.
Study design
This was a prospective, single-center diagnostic accuracy study registered on ClinicalTrials.gov (NCT05716620). The investigators enrolled patients with cirrhosis who had an annual HCC risk greater than 5%, placing them in a clearly high-risk surveillance group.
Participants underwent paired screening with both ultrasound and non-contrast AMRI in two rounds, six months apart. If either imaging test was positive, or if there was clinical suspicion of HCC, the patient then underwent multiphasic contrast-enhanced MRI (CE-MRI), which served as the reference standard for confirming cancer.
The main outcome was the HCC detection rate, measured as per-patient sensitivity. The study also assessed specificity, overall diagnostic accuracy, lesion-level detection, stage at diagnosis, and interobserver agreement between readers.
Key findings
A total of 614 paired screening examinations were performed in 404 patients. Of these, 97 patients proceeded to CE-MRI because of positive screening results or clinical concern. CE-MRI identified 39 HCC lesions in 37 patients.
The results showed a clear advantage for AMRI over ultrasound:
AMRI sensitivity was 94.6% compared with 51.4% for ultrasound.
AMRI specificity was 96.6% compared with 69.5% for ultrasound.
The area under the receiver operating characteristic curve (AUROC) was 0.956 for AMRI versus 0.604 for ultrasound.
All of these differences were statistically significant, with p<0.001.
At the lesion level, AMRI detected 37 of 39 HCC lesions (94.9%), while ultrasound detected 20 of 39 (51.3%). In practical terms, AMRI found nearly all cancers identified by the reference MRI, whereas ultrasound missed about half.
Importantly, among HCCs detected by AMRI, 97.3% were early-stage disease, classified as BCLC stage 0 or A. This is highly relevant because early-stage HCC is the stage at which curative treatments are most likely to be effective.
Reader agreement was also better for AMRI. Interobserver agreement was described as almost perfect for AMRI (κ=0.929), compared with moderate agreement for ultrasound (κ=0.631). This suggests that AMRI may be more consistent and less dependent on operator skill or patient body habitus.
Why these findings matter
The main goal of HCC surveillance is not simply to find cancer, but to find it early enough to change outcomes. When HCC is detected at an early stage, patients may be eligible for curative options such as surgical resection, liver transplantation, or local ablation. Missing early tumors can delay treatment and reduce survival.
Ultrasound has long been favored because it is practical, but its limitations are well known. Performance can vary depending on the operator, the quality of the acoustic window, the size and location of the lesion, and the background of cirrhosis itself. In many real-world settings, these weaknesses reduce its value as a surveillance tool.
The findings from this study suggest that a rapid non-contrast AMRI protocol may overcome many of these limitations while still being more feasible than a full MRI exam. Because no contrast agent was needed, the protocol may also be simpler to incorporate into surveillance pathways and may be more acceptable for patients with contraindications to contrast-enhanced studies.
Clinical interpretation
These interim results are encouraging for several reasons.
First, they support the idea that abbreviated MRI can achieve substantially better sensitivity than ultrasound in a high-risk cirrhotic population. Sensitivity is particularly important in surveillance because the main purpose is to avoid missed cancers.
Second, the high specificity of AMRI suggests that it may not create excessive false alarms. A screening test with low specificity can lead to unnecessary anxiety, extra imaging, and invasive procedures. In this study, AMRI performed well on both fronts.
Third, the fact that almost all AMRI-detected cancers were early-stage strengthens the clinical relevance of the findings. Finding more cancers is important, but finding curable cancers is even more meaningful.
That said, diagnostic accuracy alone does not prove better patient outcomes. A surveillance test must ultimately be judged by whether it improves survival, increases curative treatment rates, reduces advanced-stage presentations, and does so at an acceptable cost and burden to the health system. Those questions require longer-term prospective studies.
Limitations
Although the results are promising, several limitations should be kept in mind.
This was an interim analysis from a single-center study, so the findings may not fully generalize to all practice settings. Performance could differ in community hospitals, in populations with different causes of cirrhosis, or in centers with varying MRI expertise.
The number of confirmed HCC cases was still relatively small, which is common in surveillance studies but means the confidence intervals remain important. Larger studies will be needed to confirm the magnitude of benefit.
Also, the reference standard was multiphasic contrast-enhanced MRI, which is highly reliable but not identical to pathology in every case. In surveillance research, this is a common and accepted approach, yet it remains a methodological consideration.
Finally, the study evaluated diagnostic performance rather than downstream outcomes. It does not yet answer whether AMRI-based surveillance improves survival, is cost-effective, or should replace ultrasound across all cirrhotic patients.
Implications for practice
For now, ultrasound remains the standard surveillance tool in many guidelines because of its accessibility and low cost. However, this study adds to a growing body of evidence suggesting that abbreviated MRI may be a stronger option for selected high-risk patients, especially those in whom ultrasound quality is expected to be poor.
Potential candidates for AMRI-based surveillance may include patients with obesity, markedly heterogeneous cirrhotic livers, prior nondiagnostic ultrasounds, or other situations where ultrasound has repeatedly been limited. AMRI may also be valuable in centers with the infrastructure to support MRI surveillance programs.
If future studies confirm improved clinical outcomes, abbreviated MRI could become an important complement to or replacement for ultrasound in certain surveillance pathways. In the meantime, many experts would likely view it as an emerging option rather than a universal standard.
Conclusion
In this prospective interim surveillance trial, a rapid non-contrast abbreviated MRI protocol showed substantially better diagnostic performance than ultrasound for detecting hepatocellular carcinoma in patients with cirrhosis. AMRI had higher sensitivity, higher specificity, better interobserver agreement, and detected a very high proportion of early-stage cancers.
These findings suggest that abbreviated MRI could help close a major gap in HCC surveillance, particularly for high-risk patients in whom ultrasound is less reliable. Still, before any definitive guideline change can be recommended, larger prospective studies must show that this improved detection translates into better patient-centered outcomes such as curative treatment rates and survival.
