Serum protein induced by vitamin K absence or antagonist-II predicts aggressive tumor biology in alpha-fetoprotein-normal hepatocellular carcinoma

BACKGROUND Patients with hepatocellular carcinoma (HCC) beyond the Milan criteria or with portal vein tumor thrombosis are often excluded from the transplant list owing to aggressive biology and recurrence risk. While high alpha-fetoprotein (AFP) signals aggressiveness, the behavior of normal AFP HCC with elevated protein induced by vitamin K absence/antagonist-II (PIVKA-II) is less defined. AIM To assess the prognostic value of PIVKA-II in normal AFP HCC. METHODS Retrospective cohort of 113 patients with normal AFP and normal or elevated PIVKA-II. “Aggressive” tumors were defined as beyond Milan and/or portal vein tumor thrombosis (n = 63); others were non-aggressive (n = 50). Receiver operating characteristic curve analysis identified PIVKA-II cut-offs. RESULTS This study included 78 men and 35 women; mean age 58.4 ± 11.1 years; 62.8% with decompensated cirrhosis. PIVKA-II was higher in aggressive tumors: Median 2785 mAU/mL (interquartile range: 222-8152) vs 239 mAU/mL (interquartile range: 55-727), P < 0.001. Area under receiver operating characteristic curve 0.756 (95% confidence interval: 0.669-0.844). The Youden-optimized cut-off for aggressive HCC was 1609.5 mAU/mL [sensitivity: 0.54, specificity: 0.94; positive predictive value (PPV): 0.919]. A sensitivity-oriented cut-off of 400 mAU/mL gave sensitivity 0.69 and specificity 0.64 (PPV: 0.71). Regression analysis showed that PIVKA-II > 400 mAU/mL was strongly associated with aggressive tumor phenotype (adjusted odds ratio = 5.16, P = 0.001). All the 29 patients with > 4000 mAU/mL were in the aggressive group (PPV: 1.0). All thresholds were dataset-derived. CONCLUSION In normal AFP HCC, PIVKA-II discriminates aggressive biology. A cut-off of 1609.5 mAU/mL balances sensitivity and specificity; 400 mAU/mL favors sensitivity; > 4000 mAU/mL delineates an ultra-high-risk subgroup. Findings support the incorporation of PIVKA-II into risk stratification.