Noncontrast MRI with diffusion-weighted imaging as the sole imaging modality for detecting liver malignancy in patients with high risk for hepatocellular carcinoma

Purpose: To compare the diagnostic performance of the noncontrast MRI including DWI to the standard MRI for detecting hepatic malignancies in patients with chronic liver disease. Materials and methods: We included 135 patients with 136 histologically-confirmed hepatocellular carcinomas (HCCs), 12 cholangiocarcinomas, and 34 benign lesions (<= 2.0 cm), and 22 patients with cirrhosis but no focal liver lesion who underwent 3.0 T liver MRI. Noncontrast MRI set (T1- and T2-weighted images and DWI) and standard MRI set (gadoxetic acid-enhanced and noncontrast MRI) were analyzed independently by three observers to detect liver malignancies using receiver operating characteristic analysis. Results: The Az value of the noncontrast MRI (mean, 0.906) was not inferior to that of the combined MRI (mean, 0.924) for detecting malignancies by all observers (P > 0.05). For each observer, no significant difference was found in the sensitivity and specificity between the two MRI sets for detecting liver malignancies and distinguishing them from benign lesions (P > 0.05), whereas negative predictive value was higher with the combined MRI than with the noncontrast MRI (P = 0.0001). When using pooled data, the sensitivity of the combined MRI (mean 94.8%) was higher than that of the noncontrast MRI (mean, 91.7%) (P = 0.001), whereas specificity was equivalent (78.6% vs 77.5%). Conclusion: Noncontrast MRI including DWI showed reasonable performance compared to the combined gadoxetic acid-enhanced and noncontrast MRI set for detecting HCC and cholangiocarcinoma and differentiating them from benign lesions in patients with chronic liver disease. (C) 2014 Elsevier Inc. All rights reserved.