Enhanced diagnostic performance for subcentimeter hepatocellular carcinoma using a novel criterion integrating serum AFP levels and gadolinium-based contrast-enhanced MRI features.

Purpose: To assess the effectiveness of LI-RADS v2018 and r-LI-RADS in diagnosing subcentimeter hepatocellular carcinoma (HCC) and to evaluate the potential value of serum alpha-fetoprotein (AFP) in conjunction with gadolinium-based contrast-enhanced MRI (CE-MRI) for assessing these lesions.

Methods: This retrospective study included 179 untreated, high-risk patients with microlesions (< 1 cm) from 2015 to 2023. Of these, 92 lesions were pathologically confirmed as HCC, the remaining 87 were non-HCC. Two radiologists independently rated imaging features using LI-RADS and r-LI-RADS. The optimal AFP threshold for HCC diagnosis was determined by the Youden index. Logistic regression analyses identified independent predictors of micro-HCC, leading to the development of new diagnostic criteria.

Results: Multivariate analysis identified AFP > 12.15 ng/mL, non-peripheral arterial phase enhancement, diffusion restriction, fat deposition, and enhancing capsule as key independent factors for diagnosing micro-HCC. A new criterion requiring at least three positive factors improved sensitivity to 78.3% vs. 58.7% for LR-4 (p = 0.001), with similar specificity (83.9% vs. 81.6%, p = 0.824). It also outperformed r-LR-5/4 and r-LR-4 in sensitivity (78.3% vs. 59.8% and 58.7%, both p = 0.001), without impacting specificity.

Conclusion: A new criterion (at least three positive findings among AFP > 12.15 ng/mL, non-peripheral arterial phase enhancement, diffusion restriction, fat deposition, and enhancing capsule) significantly improves the diagnostic sensitivity for subcentimeter HCC while maintaining high specificity, demonstrating clear advantages over the LR-4, r-LR-5/4, and r-LR-4 standards of LI-RADS.