Qualitative and quantitative assessment of gadoxetic acid MRI in distinguishing atypical focal nodular hyperplasia from hepatocellular adenoma subtypes
摘要
To assess qualitative and quantitative imaging features, including lesion-to-liver contrast enhancement ratio (LLCER), on gadoxetic acid-enhanced MRI for differentiating atypical focal nodular hyperplasia (aFNH) from hepatocellular adenoma (HCA) subtypes.
Materials and methodsThis retrospective study included patients with histopathologically-confirmed aFNH and HCA who underwent gadoxetic acid-enhanced MRI between January 2010 and December 2020. Two radiologists assessed qualitative imaging features and calculated LLCERs for quantitative evaluation of HBP enhancement. Statistical analyses included ROC curves, sensitivity, specificity, and a decision tree.
Results86 patients (41 ± 11 years; 64 women) had 29 aFNHs and 90 HCAs. HBP iso-/hyperintensity was observed in 72.4% (21/29) of aFNH compared to 28.8% (15/52) of U-HCA, 35% (7/20) of I-HCA, and 0% (0/11) of H-HCA. β-HCA showed 71.4% (5/7) iso-/hyperintensity, overlapping with aFNH (p = 0.17). Homogeneous iso-/hyperintensity and rim-like enhancement were present in 48.3% (14/29) and 20.7% (6/29) of aFNH but absent in β-HCA (p = 0.004). LLCER demonstrated high diagnostic performance, differentiating aFNH from U-HCA, H-HCA, and I-HCA(AUCs 0.91–0.99, sensitivities 82.8–96.6%, specificities 90.0–100%). For β-HCA, LLCER showed overlap with aFNH; AUCs (0.62–0.64) and specificities (57.1%). Among HCAs with HBP iso-/hyperintensity, 74.1% had negative LLCER values, while 25.9% (mostly β-HCA/U-HCA) showed positive values (true enhancement). Combined decision trees selected LLCER as the primary node, with downstream HBP features variably incorporated, except β-HCA, which solely used qualitative features.
ConclusionHBP iso-/hyperintensity and LLCER reliably differentiated aFNH from most HCA subtypes, except β-HCA, validated by decision tree analyses. Homogeneous iso-/hyperintensity or rim-like enhancement was absent in β-HCA, aiding differentiation.
Key Points