Detecting hepatocellular carcinoma: value of unenhanced or arterial phase CT imaging or both used in conjunction with conventional portal venous phase contrast-enhanced CT imaging

Because rates of detection of hypervascular neoplasms by conventional dynamic incremental-bolus CT are lower than rates of detection of hypovascular tumors by CT and because both unenhanced CT imaging and arterial phase helical CT imaging may increase the detection of hypervascular tumors, such as hepatocellular carcinoma, we evaluated the value of unenhanced and arterial phase CT imaging used in conjunction with conventional portal venous phase CT imaging in patients with hepatocellular carcinoma. Unenhanced and biphasic helical contrast-enhanced CT studies were performed on 81 patients with proven hepatocellular carcinoma. Arterial phase and portal venous phase images were obtained at 20-50 sec and at 60-100 sec, respectively. Three blinded readers evaluated portal venous phase images for the number of liver lesions. On separate dates, the readers compared the arterial phase images with the portal venous phase images and the unenhanced images with the portal venous phase images. The readers recorded the number of lesions that were seen on portal venous phase images and that were also detected on unenhanced or arterial phase images as well as the number of additional lesions seen on unenhanced or arterial phase images. Consensus readings of unenhanced, arterial phase, and portal venous phase images were obtained in the 42 patients who had definitive surgery or follow-up CT scans, documenting the total tumor burden in this patient subgroup. The readers identified 286-310 lesions on portal venous phase images. On unenhanced images, the readers identified 223-244 of the lesions seen on portal venous phase images and an additional 45-55 lesions that were not seen on portal venous phase images. Arterial phase imaging revealed 245-269 of the lesions seen on portal venous phase images and an additional 89-111 lesions that were not seen on portal venous images. The diagnosis of tumor was possible only on unenhanced images in two (3%) of 81 patients and only on arterial phase images in seven patients (9%). In the subset of 42 patients with proof of tumor burden, 157 proven lesions were found. Consensus readings identified 127 (81%) of these lesions on portal venous phase images, 98 (62%) of these lesions on unenhanced images, and 120 (76%) of these lesions on arterial phase images. Of the 30 lesions not seen on portal venous phase images, nine were seen on both unenhanced and arterial phase images, three were seen on unenhanced images only, and 18 were seen on art