Thorsten Persigehl MD, Presenter: Nothing to Disclose

Xiaotao Guo PHD, Abstract Co-Author: Nothing to Disclose

Elizabeth Verna, Abstract Co-Author: Nothing to Disclose

Jean Emond, Abstract Co-Author: Nothing to Disclose

Lawrence H. Schwartz MD, Abstract Co-Author: Nothing to Disclose

Binsheng Zhao DSc, Abstract Co-Author: License agreement, Varian Medical Systems, Inc License agreement, Keosys License agreement, Hinacom Software and Technology, Ltd License agreement, AG Mednet, Inc

Liver transplantation (LT) represents the only curative treatment of HCC in liver cirrhosis. Commonly used morphologic selection criteria, such as Milan criteria, are mainly based on tumor size and number of lesions, but do not take into account microvascular invasion as a major risk factor for tumor recurrence after LT. The purpose of this pilot study was to evaluate semi-automatic imaging feature analysis for assessment of vascular invasion at first staging MRI of HCC patients who underwent LT.

In this IRB-approved, retrospective pilot study, baseline MRIs (from 2003-2009) of 88 HCC patients with a total of 144 suspicious lesions were included. Lesions were semi-automatically delineated at arterial DCE-MRI. The imaging features of 2D roundness factor (RF: defined as a function of tumor perimeter and area) and 3D compactness factor (CF: defined as a function of tumor surface and volume), as well as the maximum tumor diameter (Dia) and volume (Vol) were calculated. Computer-derived results were correlated (A) for all HCC lesions and (B) for the worst index lesion per patient (e.g. lowest RF) with the pathologically reported micro- and/or macrovascular invasion at the explanted liver after LT (from 2004-2010). Chi-square (p-value) and AIC-based statistics were calculated.

Despite general limitations (e.g. various bridging times and different treatments before LT), we found a positive association between all imaging findings at staging MRI (RF,CF,Dia,and Vol) and any vascular HCC invasion at explant pathology after LT (p=.001/.009/.003/.019). However, the combined RF/Dia performed best (RF/Dia=88.9;RF=89.9, and Dia=93.7). Moreover, RF and CF correlated strongly with the microvascular invasion on lesion (p=.002/.009) and patient basis (p=.003/.045).

In our pilot HCC study, the semi-automatic calculated roundness factor (RF) seemed to allow a non-invasive prediction of vascular invasion at the staging MRI, and performed better than simple size or volume measurements.

Non-invasive semi-automatic imaging feature analysis might provide an independent staging biomarker for new advanced selection criteria in HCC before liver transplantation.

Persigehl, T, Guo, X, Verna, E, Emond, J, Schwartz, L, Zhao, B, Semi-automatic Imaging Feature Analysis for Assessment of Vascular Invasion in Hepatocellular Carcinoma (HCC): A Radiogenomic Pilot Study.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14013085.html