New HCC Blood Test HES V2.0 Outperforms Rivals in Early Liver Cancer Detection

Hepatocellular carcinoma (HCC) remains one of the deadliest cancers globally, with 5-year survival below 20% unless caught early enough for curative treatment. Current surveillance for high-risk patients—primarily those with cirrhosis—relies on liver ultrasound plus alpha-fetoprotein (AFP), a combination with well-documented sensitivity limitations. More sophisticated biomarker panels like GALAD (incorporating gender, age, AFP-L3, AFP, and DCP) have shown promise, but head-to-head external validation studies in prospective cohorts are scarce.

This study externally validated HES V2.0, a biomarker algorithm incorporating AFP, AFP-L3%, DCP, age, platelets, ALT, cirrhosis etiology, and critically, the one-year rate of change (gradient) for AFP, AFP-L3, and DCP. The validation was conducted in the HEDS cohort—a rigorously designed, multicenter NIH-sponsored prospective study enrolling 1,485 cirrhosis patients across seven US academic centers between 2013 and 2021. Of these, 119 developed HCC during 5,008 person-years of follow-up. Stored serum samples were tested retrospectively in blinded fashion on the Wako i30 immunoanalyzer, following PRoBE phase III biomarker trial methodology to minimize bias.

At a fixed 10% false positive rate (90% specificity), HES V2.0 demonstrated significantly higher patient-level sensitivity than GALAD at 12 months before HCC diagnosis (53.7% vs. 41.8%, p=0.025) and trended higher at 6 and 24 months. Against ASAP, HES V2.0 was superior by 10.9–16.3% across all time windows (p<0.01 at 12 months, 24 months, and anytime before diagnosis). Compared to its predecessor HES V1.0, HES V2.0 achieved 11.9% higher sensitivity at 12 months (p=0.007). Both HES V2.0 and GALAD shared an identical overall AUROC of 0.79, indicating equivalent global discrimination but importantly different sensitivity-specificity tradeoff profiles.

A particularly notable finding emerged when analyses were restricted to patients with at least two longitudinal biomarker measurements enabling gradient calculation: HES V2.0 outperformed GALAD by 3.5–8.7% at all time points, representing an 8.7–24.0% relative increase in sensitivity, with several comparisons reaching statistical significance. This underscores that the gradient-incorporating feature of HES V2.0 is a genuine driver of its advantage, not just a statistical artifact. At the GALAD-specific threshold (18.1% FPR), however, GALAD slightly outperformed HES V2.0 at 6 and 12 months, suggesting the two scores have complementary strengths depending on the operating threshold chosen.

Most HCC cases met Milan criteria (71.2%) and 42.8% were BCLC stage 0/A, confirming the cohort captures clinically meaningful early-stage disease. The study's external validation design, prospective sample collection, blinded assaying, and multicenter enrollment substantially strengthen confidence in these findings compared to internal cross-validation approaches. The results support HES V2.0 as a strong candidate for integration into HCC surveillance programs, particularly where serial blood draws every 6 months are already standard practice.