Hepatocellular carcinoma (HCC) surveillance with semiannual ultrasound and AFP is guideline-recommended for patients with cirrhosis and selected patients with chronic hepatitis B. However, this strategy has suboptimal sensitivity for early-stage HCC and poor real-world adherence; moreover, randomized trial evidence demonstrating reduced HCC-related or all-cause mortality in patients with cirrhosis is lacking. This Expert Opinion evaluates two emerging categories of HCC surveillance tests - abbreviated magnetic resonance imaging (aMRI) protocols and blood-based biomarker panels - and proposes that, if validated, their future clinical implementation should be guided not only by HCC risk, but by expected surveillance benefit. Protocols for aMRI seek to preserve the high diagnostic accuracy of full liver-protocol MRI while reducing scan time, cost, and patient burden. Dynamic contrast-enhanced aMRI (DCE-aMRI) can establish a definitive HCC diagnosis without recall imaging and may match the accuracy of full liver-protocol MRI. Non-contrast aMRI (NC-aMRI) avoids intravenous contrast and has demonstrated superior sensitivity and specificity compared with ultrasound in recent trials. Blood-based biomarker panels, including protein-based tests (GALAD, HES 2.0) and circulating tumor DNA-based assays (Oncoguard, HelioLiver) may improve adherence, but face methodological challenges and require rigorous, prospective longitudinal evaluation to determine whether they result not only in earlier actionable detection but also better clinical outcomes. We argue that selecting which patients should receive enhanced surveillance requires moving beyond risk stratification toward benefit stratification, i.e., towards basing surveillance recommendations not only on the patient's risk of developing HCC, but also on the probability that by enabling earlier-stage detection and receipt of effective treatment, surveillance improves clinical outcomes. Patients at the highest HCC risk do not necessarily stand to gain the most from earlier detection, underscoring the need for pragmatic frameworks and prediction tools that estimate individualized surveillance benefit.
