Stephanie Channual MD, Presenter: Nothing to Disclose

Anokh Pahwa MD, Abstract Co-Author: Nothing to Disclose

Katrina Richards Beckett MD, Abstract Co-Author: Nothing to Disclose

James Sayre PhD, Abstract Co-Author: Nothing to Disclose

David Shin-Kuo Lu MD, Abstract Co-Author: Consultant, Covidien AG Speaker, Covidien AG Consultant, Johnson & Johnson Research Grant, Johnson & Johnson Consultant, Bayer AG Research Grant, Bayer AG Speaker, Bayer AG

Steven Satish Raman MD, Abstract Co-Author: Consultant, Bayer AG Consultant, Covidien AG

Only recently has LI-RADS (LR) expanded to apply to hepatobiliary (HB) contrast agents, with lesion appearance on the HB phase considered to be an ancillary feature that favors the diagnosis of hepatocellular carcinoma (HCC). In contrast, the Japan Society of Hepatology (JSH) includes lesion appearance on the HB phase as a major criteria that favors the diagnosis of HCC. The purpose of our study was to determine the performance of LI-RADS v2014 and Japan Society of Hepatology (JSH) 2010 criteria for the non- invasive diagnosis of HCC.

This was an IRB approved, HIPAA compliant retrospective study with 131 consecutive suspected HCC nodules in 114 patients confirmed by percutaneous biopsy, resection, or explant within 90 days of Gd-EOB-DTPA MRI. Nodule size, presence of a capsule, and enhancement patterns were recorded. The nodules were then categorized as LR3, LR4, or LR5 based on the LI-RADS major criteria, and categorized as either meeting or not meeting the JSH criteria (defined as arterial enhancement and venous wash out, or arterial enhancement and lack of Gd-EOB-DTPA uptake on HB phase imaging).

Of the 131 nodules, 116 were pathologically confirmed HCC (88.5%). Of 131 nodules, 23 (18%), 41 (31%), and 67 (51%) were categorized as LR3, LR4, and LR5 respectively. Of these, 15/23, 37/41, and 64/67 LR3, LR4 and LR5 nodules were pathologically proven as HCC, respectively (sensitivities, 13%, 32%, and 55%, respectively; specificities, 47%, 73%, and 80%, respectively). The PPV of LR3, LR4, and LR5 were 65%, 90%, and 96%, respectively. The sensitivity, specificity, and PPV for the JSH criteria were 72.4%, 53.3%, and 92.3%, respectively. The accuracy of LR4 and LR5 combined was 83% (109/131), while the accuracy for the JSH criteria was 70.2% (92/131).

Although use of LI-RADS with Gd-EOB-DTPA yields a high PPV and accuracy for diagnosing HCC, moderate sensitivity and specificity suggest that further refinement of the criteria may be necessary and percutaneous nodule biopsy may be complementary for diagnosis. However, LR4 and LR5 combined was more sensitive and accurate for diagnosing HCC compared to the JSH criteria.

The use of hepatobiliary specific MR contrast agents, such as Gd-EOB-DTPA, is becoming more prevalent, and understanding its applicability with LI-RADS is essential for the noninvasive evaluation of nodules in cirrhotic livers.

Channual, S, Pahwa, A, Beckett, K, Sayre, J, Lu, D, Raman, S, Performance of LI-RADS Criteria for Diagnosis of Pathologically Proven Hepatocellular Carcinoma Using Gd-EOB-DTPA, and Comparisons with the Japan Society of Hepatology 2010 Criteria.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14016148.html