Enhancing non-LR-5 liver lesion diagnoses: the incremental value of serum alpha-fetoprotein and PIVKA-II for detecting hepatocellular carcinoma components

Backgrounds/Aims: To investigate whether incorporating serum alpha-fetoprotein (AFP) and protein induced by vitamin K absence or antagonist-II (PIVKA-II) into gadoxetic acid-enhanced liver MRI improves the accuracy of differentiating hepatocellular carcinoma (HCC), intrahepatic cholangiocarcinoma (ICC), and combined hepatocellular-cholangiocarcinoma (cHCC-CC) from non-LR-5 lesions. Methods: This retrospective study included 123 patients comprising 70 HCC, 30 ICC, and 23 cHCC-CC cases who underwent preoperative MRI and tumor marker analysis. Two abdominal radiologists independently interpreted the MRI without markers (Decision 1). Then, lesions were reclassified as HCC if either marker was elevated, regardless of the initial interpretation (Decision 2, biomarker-assisted override strategy). Outcomes included the diagnostic performance for identifying HCC versus non-HCC malignancies and distinguishing the presence of HCC components (HCC or cHCC-CC) from ICC. The McNemar's test and Cohen's κ compared diagnostic performance and evaluated interobserver agreement, respectively. Results: Adding AFP and PIVKA-II (Decision 2) improved the sensitivity and accuracy of detecting HCC components from 62.4-76.3% to 81.7-90.3% (P<0.001) and 66.7-76.4% to 81.3-87.0% (P<0.001), respectively. The overall accuracy for differentiating HCC from non-HCC malignancies did not improve. However, the accuracy for detecting HCC in small tumors (<3 cm) significantly improved from 68.8-72.5% to 81.3-82.5% (P= 0.002-0.008). Interobserver agreement also increased, from moderate (κ = 0.48) to substantial (κ = 0.68) after adding tumor marker values. Conclusions: In non-LR-5 hepatic lesions, adding AFP and PIVKA-II to MRI interpretations significantly improved the accuracy of identifying HCC components, as well as HCC detection in small tumors.