Efficacy of automated tumor-feeder detection software using cone-beam computed tomography technology in transarterial embolization through extrahepatic collateral vessels for malignant hepatic tumors.

Abstract

We evaluated the performance of automated tumor-feeder detection (AFD) software using cone-beam computed tomography technology in identifying tumor-feeders of extrahepatic collaterals. AFD was prospectively used in superselective transarterial chemoembolization (TACE) or embolization (TAE) of extrahepatic collaterals for 29 hepatocellular carcinomas and one liver metastasis (mean tumor diameter ± standard deviation, 28 ± 15.6 mm) in 25 patients. The detectability of extrahepatic tumor-feeders with non-selective digital subtraction angiography (DSA) and AFD was evaluated and compared using a χ(2) -test. Tumor response of target lesions in each patient at 2-3 months after treatment was evaluated using the modified Response Evaluation Criteria in Solid Tumors. Complications were also evaluated. Of 46 tumor-feeders, non-selective DSA and AFD could identify 26 and 44, respectively (P < 0.001). Regarding the origin of tumor-feeders, both non-selective DSA and AFD could identify 14 of 15, six of seven and two of two tumor-feeders of the right inferior phrenic, omental and right renal capsular artery, respectively. In the cystic and left gastric or right colic artery, AFD could identify 13 of 13 and nine of nine tumor-feeders but non-selective DSA could identify only three of 13 and one of nine, respectively (P < 0.001). Complete response was obtained in 15 patients, partial response in six, stable disease in three and progressive disease in one. No severe complications developed except for right pleural effusion after embolization of the right inferior phrenic artery. AFD has a sufficient ability to identify extrahepatic tumor-feeders and may improve the safety and local effects of TACE/TAE of extrahepatic collaterals.