Sang Ho Lee PhD, Presenter: Nothing to Disclose

Koichi Hayano MD, Abstract Co-Author: Nothing to Disclose

Dushyant V. Sahani MD, Abstract Co-Author: Research Grant, General Electric Company

Andrew X. Zhu MD, PhD, Abstract Co-Author: Nothing to Disclose

Hiroyuki Yoshida PhD, Abstract Co-Author: Patent holder, Hologic, Inc Patent holder, MEDIAN Technologies

To compare five different standard dual-input pharmacokinetic models (PKMs) with the fast water exchange regime for the analysis of baseline DCE-MRI data in the prediction of 1-year survival (1YS) and its association with overall survival (OS) in advanced hepatocellular carcinoma (HCC) treated by sunitinib.

Twenty patients with advanced HCC underwent DCE-MRI at baseline, and received sunitinib daily by mouth for 28 days followed by 14 days of rest in 6-week cycles. The baseline DCE-MRI data were analyzed retrospectively by using five different standard dual-input PKMs: the Tofts-Kety (TK), extended TK, two compartment exchange, adiabatic approximation to the tissue homogeneity (AATH), and distributed parameter (DP) models. Kinetic parameters consisted of total hepatic blood flow (BF), arterial flow fraction (γ), arterial BF (BFA), portal BF, blood volume, mean transit time, capillary permeability-surface area product (PS), fractional interstitial volume (vI), and extraction fraction (E). Following receiver operating characteristic analysis with additional leave-one-out cross-validation, parameters of the different kinetic models were compared in terms of 1YS discrimination using cross-validated Kaplan-Meier analysis, and association with OS using a univariate Cox-proportional hazard model, with additional permutation testing.

For 1YS prediction, the TK-model-derived vI (P=0.037), the AATH-model-derived BFA (P=0.019), PS (P=0.027), and E (P=0.033), and the DP-model-derived γ (P=0.012) and BFA (P=0.041) had statistically significant predictability after cross-validation and permutation testing, all of which were lower in the high-risk group. For OS, the increase of the AATH-model-derived PS and the DP-model-derived BFA were statistically significantly associated with the increase of OS with hazard ratios of 0.766 (P=0.023) and 0.809 (P=0.025) after permutation testing, respectively.

The AATH-model-derived PS and the DP-model-derived BFA were effective biomarkers for both the prediction of 1YS and the association with OS. Among the standard models, the AATH and DP were favorable models in survival analysis.

Kinetic parameters derived from dual-input PKMs with the fast water exchange regime based on baseline DCE-MRI data can provide effective prognostic imaging biomarker.

Lee, S, Hayano, K, Sahani, D, Zhu, A, Yoshida, H, DCE-MRI-based Pharmacokinetic Biomarker for Predicting Survival of Patients with Advanced Hepatocellular Carcinoma Treated by Sunitinib: Fast-Water-Exchange-Limit-Constrained Analysis.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14005837.html