Purpose: Response Evaluation Criteria in Solid Tumors (RECIST) may underestimate the efficacy of targeted therapies. In hepatocellular carcinoma (HCC) studies with sunitinib, RECIST-defined response rates are low, although hypodensity on computed tomography (CT) scans occurs more frequently. This exploratory analysis investigated tumor density as a surrogate endpoint of sunitinib activity in a phase II HCC study.Experimental Design: Patients received sunitinib 50 mg/d (4 weeks on/2 weeks off). Tumor size and density were assessed on CT scans by using RECIST and Choi criteria, the latter of which classify a partial response as a 15% or more reduction in tumor density or a 10% or more reduction in tumor size. The overall percentage volume of tumor necrosis was calculated with volumetric reconstruction. Tumor perfusion parameters were assessed by using perfusion CT scans with specific acquisition.Results: Among the 26 evaluable patients, 1 achieved a partial response and 22 had tumor stabilization by RECIST. In analysis of tumor density, 17 of 26 patients (65.4%) were responders by Choi criteria. Volumetric assessment showed major tumor necrosis (!30% of tumor volume) in 10 of 21 patients (47.6%). Among four patients evaluated, tumor blood flow was reduced by 58.8% and blood volume by 68.4% after 4 weeks of treatment. The median time to progression (TTP) was 6.4 months. Patients with responses by Choi criteria had a significantly longer TTP (7.5 months) compared with nonresponders (4.8 months; HR ¼ 0.33, two-sided P ¼ 0.0182).Conclusions: Tumor density assessment suggested that radiologic endpoints in addition to RECIST may be considered to capture sunitinib activity in HCC. Clin Cancer Res; 17(13); 4504-12. Ó2011 AACR.