Abstract
OBJECTIVE: To validate the diffusion-weighted MRI (DWI) for differentiation of benign from malignant portal vein thrombosis. MATERIALS AND METHODS: The Institutional Review Board approved this retrospective study and waived informed consent. A total of 59 consecutive patients (52 men and 7 women, aged 40-85 years) with grossly defined portal vein thrombus (PVT) on hepatic MRI were retrospectively analyzed. Among them, liver cirrhosis was found in 45 patients, and hepatocellular carcinoma in 47 patients. DWI was performed using b values of 50 and 800 sec/mm(2) at 1.5-T unit. A thrombus was considered malignant if it enhanced on dynamic CT or MRI; otherwise, it was considered bland. There were 18 bland thrombi and 49 malignant thrombi in 59 patients, including 8 patients with simultaneous benign and malignant PVT. Mean apparent diffusion coefficients (ADCs) of benign and malignant PVTs were compared by using Mann-Whitney U test. Diagnostic accuracy was evaluated using receiver operating characteristic (ROC) curve analysis. RESULTS: The mean ADC ± standard deviation of bland and malignant PVT were 1.00 ± 0.39 × 10(-3) mm(2)/sec and 0.92 ± 0.25 × 10(-3) mm(2)/sec, respectively; without significant difference (p = 0.799). The area under ROC curve for ADC was 0.520. An ADC value of > 1.35 × 10(-3) mm(2)/sec predicted bland PVT with a specificity of 94.6% (95% confidence interval [CI]: 84.9-98.9%) and a sensitivity of 22.2% (95% CI: 6.4-47.6%), respectively. CONCLUSION: Due to the wide range and considerable overlap of the ADCs, DWI cannot differentiate the benign from malignant thrombi efficiently.