Return To Abstract Listing

Scientific Session 24 — Gastrointestinal - Focal Liver and Biliary Disease

Thursday, May 4, 2017

Abstracts 2218-3267



2423. LI-RADS Ancillary Features of Malignancy on MRI in Patients at Risk for Hepatocellular Carcinoma

Cerny M1*,  Bergeron C1,  Olivié D1,  Murphy-Lavallée J1,  Tang A1,2 1. Centre hospitalier de l'Université de Montréal, Montreal, Canada; 2. Centre de recherche du Centre hospitalier de l'Université de Montréal, Montreal, Canada

Address correspondence to M. Cerny (milenacerny@gmail.com)

Objective: The purpose of this study was to assess the impact of Liver Imaging Reporting and Data System (LI-RADS) MRI ancillary features of malignancy on observation category and sensitivity for the diagnosis of hepatocellular carcinoma (HCC).

Materials and Methods: This retrospective study was approved by our institutional review board. Patient consent was waived. Over a 7-month interval, 67 patients with suspected HCC underwent diagnostic MRI with extracellular contrast agents. When applicable, the presence or absence of major and ancillary features was assessed for 155 observations and assigned a LI-RADS category. The composite reference standard was histopathology (n = 51), follow-up imaging after at least 6 months (n = 79), or presumptive diagnosis (LIRADS category 4 or 5) based on major features (n = 25). Frequency of category changes and per-lesion sensitivity of ancillary features of malignancy were estimated. Sensitivity of ancillary features suggestive of malignancy was calculated.

Results: Ancillary features modified the final LI-RADS category of 11/155 (7.0%) observations. The final category was upgraded in 2/155 (1.3%) and downgraded in 9/155 (5.8%) of observations. For the diagnosis of HCC, the sensitivity of individual ancillary features was 41.3% (26/63) for subthreshold growth, 26.8% (40/149) for intralesional fat, 24.3% (9/37) for lesional fat sparing, 18.2% (2/11) for lesional iron sparing, 0.7% (1/147) for nodule-in-nodule architecture, 3.4% (5/147) for mosaic architecture, 2.7% (4/147) for corona enhancement, 12.0% (17/142) for distinctive rim, 5.3% (7/131) for blood products, 54.8% (63/115) for restricted diffusion, and 55.0% (83/151) for mild-to-moderate T2 hyperintensity. For the diagnosis of HCC, the aggregate sensitivity was 67.7% (105/155) if any ancillary feature of malignancy was present.

Conclusion: The sensitivity of individual ancillary features of malignancy in LI-RADS is modest, but the aggregate sensitivity increases if any feature is considered.