Predict microvascular invasion in hepatocellular carcinoma by dynamic contrast-enhanced magnetic resonance imaging in patients with hepatitis B virus

Purpose: To evaluate diagnostic performance of dynamic contrast-enhanced magnetic resonance imaging (MRI) for predicting microvascular invasion (MVI) of hepatocellular carcinoma (HCC) in patients with hepatitis B virus. Materials and Methods: This study was approved by our institutional review board, and the informed consent was waived. 107 patients accepted MR scanning before operation. Subsequently, 106 patients were performed liver mass resection, and one with liver transplantation. All images were evaluated by two radiologists with at least 9 years experience in abdominal MR imaging. The imaging features of tumors were as follows: peritumoral arterial enhancement, tumor pseudo-capsule, tumor margins, as well as multifocality. MVI, tumor size, Edmondson-Steiner grade and liver cirrhosis were also included. The independent Student’s t-test was used to compare continuous variables, and the chi-squared test and Fisher’s Exact test was used to analyze the categorical variables. Parameters shown to be statistically significant in univariate analysis were entered into a logistic regression model to identify independent predictors of MVI. The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated. A p-value less than 0.05 was considered to be statistically significant. Results: By histopathologic analysis, 31.8% (34/107) of the patients with HCC were confirmed as MVI-positive, and 68.2% (73/107) were as MVI-negative. MVI was present in 11.8% (4/34) of tumors less than 5 cm with smooth margins; and 68.4% (13/19) of tumors larger than 5 cm with non-smooth margins. A larger (> 5 cm) tumor size (odds ratio [OR] 4.77, 95% CI 1.50-15.17; P < 0.05) and a non-smooth MRI tumor margin (OR 3.79, 95% CI 1.07-13.43; P < 0.05) were shown as independent predictors of MVI. The sensitivity, specificity, PPV, and NPV for predicting MVI by tumor size > 5 cm were 41.2% (95% CI 25.1-59.2), 86.3% (95% CI 75.8-92.9), 58.3% (95% CI 36.9-77.2), and 75.9% (95% CI 65.0-84.3), respectively. The sensitivity, specificity, PPV, and NPV for predicting MVI by a non-smooth margin were 87.1% (95% CI 69.2-95.8), 48.6% (95% CI 36.6-60.7), 42.9% (95% CI 30.7-55.9), and 89.5% (95% CI 74.3-96.6), respectively. When combining the tumor size (> 5 cm) with tumor margin (non-smooth margin) as the risk factors, the area under ROC curve was 0.732 (95% CI 0.624-0.840), and the OR value was 4.11 (95% CI 1.99-8.48; P < 0.05) for predicting presence of MVI. Conclusion: a non-smooth tumor margin and a larger tumor size could independently predict the presence of MVI. A tumor size > 5 cm and a non-smooth tumor margin could be used as preoperative predictors of the presence of MVI in patients with HCC.