Diffusion-weighted MRI as a screening tool for hepatocellular carcinoma in cirrhotic livers: correlation with explant...

The purpose of this study was to compare the sensitivity and specificity of diffusion-weighted liver MRI alone with complete, multiphasic gadoteridol-enhanced MRI for the detection of hepatocellular carcinoma in cirrhotic patients before liver transplant.

This single institution retrospective study was performed after IRB approval and was HIPAA compliant. MRI scans of 37 patients who underwent liver transplant were evaluated and findings correlated with liver explant (36) or biopsy (1). All MRI scans were obtained within six months of explant. MRI from 17 patients with liver lesions by report at imaging subsequently proven to be HCC at pathology and 20 controls without liver lesions by imaging and pathology were reviewed in random order on the radiology PACS by three independent readers blinded to the MRI reports and pathology reports in two separate sittings. First, only the diffusion-weighted images (DWI) were interpreted. Second, the complete multiphasic MRI exam with DWI was reviewed. A consensus read was obtained by two separate radiologists who had access to the patients' explant data in order to map lesions. Reader-specific and pooled classification was assessed using sensitivity, specificity, positive predictive value, and negative predictive values and corresponding 95% confidence intervals (CI) for both DWI and complete MRI examination readings compared to pathology. McNemar's test and Kappa coefficient were used to assess differences (agreement) in DWI and complete examination readings.

A total of 37 patients have been studied (25M 12F age range 21-70). Averaged results of the three independent readers demonstrated a sensitivity of 78% (95% CI 65-89%) and specificity of 88% (95% CI 77-95%) for DWI alone for detection of liver lesions, with a positive predictive value of 85% (95% CI 72-94%) and a negative predictive value of 83% (95% CI 71-91%). Review of the complete MRI exam showed a sensitivity of 90% (95% CI 76-97%) and a specificity of 82% (95% CI 66-92%) with a positive predictive value of 83% (95% CI 69-93%) and a negative predictive value of 89% (95% CI 74-97%). McNemar's agreement test revealed no significant difference between the DWI and complete multiphasic interpretations (p = 0.3458), with simple Kappa coefficient of 0.6716 (95% CI 0.5332-0.8110). Lesions identified on DWI ranged in size from 1.5 to 5 cm. Detection of lesions was decreased in the presence of artifact from motion, large ascites, and technical issues.

Diffusion-weighted MRI has NPV and PPV comparable to complete multiphasic MRI examination for liver lesion detection in cirrhotic patients and may have a role in screening.