New hallmark of hepatocellular carcinoma, early hepatocellular carcinoma and high-grade dysplastic nodules on Gd-EOB-DTPA MRI in patients with cirrhosis: a new diagnostic algorithm.

Objective: Many improvements have been made in diagnosing hepatocellular carcinoma (HCC), but the radiological hallmarks of HCC have remained the same for many years. We prospectively evaluated the imaging criteria of HCC, early HCC and high-grade dysplastic nodules (HGDNs) in patients under surveillance for chronic liver disease, using gadolinium-ethoxybenzyl-diethylenetriamine pentaacetic acid (Gd-EOB-DTPA) MRI and diffusion-weighted imaging.

Design: Our study population included 420 nodules >1 cm in 228 patients. The MRI findings of each nodule were collected in all sequences/phases. The diagnosis of HCC was made according to the American Association for the Study of Liver Diseases (AASLD) criteria; all atypical nodules were diagnosed using histology.

Results: A classification and regression tree was developed using three MRI findings which were independently significant correlated variables for early HCC/HCC, and the best sequence of their application in a new diagnostic algorithm (hepatobiliary hypointensity, arterial hyperintensity and diffusion restriction) was suggested. This algorithm demonstrated, both in the entire study population and for nodules ≤2 cm, higher sensitivity (96% [95% CI 93.5% to 97.6%] and 96.6% [95% CI 93.9% to 98.5%], P<0.001, respectively) and slightly lower specificity (91.8% [95% CI 88.6% to 94.1%], P=0.063, and 92.7% [95% CI 88.9% to 95.4%], P=0.125, respectively) than those of the AASLD criteria. Our new diagnostic algorithm also showed a very high sensitivity (94.7%; 95% CI 92% to 96.6%) and specificity (99.3%; 95% CI 97.7% to 99.8%) in classifying HGDN.

Conclusion: Our new diagnostic algorithm demonstrated significantly higher sensitivity and comparable specificity than those of the AASLD imaging criteria for HCC in patients with cirrhosis evaluated using Gd-EOB-DTPA MRI, even for lesions ≤2 cm. Moreover, this diagnostic algorithm allowed evaluating other lesions which could arise in a cirrhotic liver, such as early HCC and HGDN.