Refining MRI-based criteria for portal vein invasion in hepatocellular carcinoma: improving sensitivity beyond portal vein tumor thrombosis

Purpose To investigate the imaging features indicating portal vein invasion (PVI) of hepatocellular carcinoma (HCC) on gadoxetic acid-enhanced MRI and to create more accurate diagnostic criteria than the presence of portal vein tumor thrombosis (PVTT) on MRI. Methods This retrospective study included patients with surgically resected HCC larger than 5 cm, and the presence of PVI was investigated. On MRI, we evaluated the image findings of portal vein occlusion, the parenchymal signal change caused by hemodynamic alterations of the portal vein, and their combination showing the highest odds ratio (OR) to define the diagnostic criteria for radiological PVI detection (rPVI criteria). The diagnostic performance and recurrence-free survival were compared between the rPVI criteria and the presence of PVTT using McNemar’s test and Kaplan–Meier method, respectively. Interobserver agreement was evaluated using Cohen’s weighted ĸ statistics. Results Of 189 enrolled patients, 25 (13.2%) had PVI on histology. To diagnose PVI on MRI, either peripheral wedge-shaped arterial peritumoral hyperemia with an abrupt cut-off of a portal vein or the presence of PVTT had the highest OR (41.67, p < 0.001). The sensitivity of PVI was significantly increased under this diagnostic criterion (64.0% to 88.0%; p = 0.031) with comparable accuracy (95.2% vs. 94.7%; p > 0.999). In terms of recurrence-free survival, the patient group with rPVI was significantly worse ( p = 0.017) compared with the patients without rPVI. Interobserver agreement of radiologic findings was substantial ( ĸ = 0.64). Conclusion Diagnostic criteria for radiologically PVI detection increase the sensitivity more than the only presence of PVTT. Graphical abstract