[Multiple-phase spiral computerized tomography of small hepatocellular carcinoma: technique optimization and diagnostic yield].

PURPOSE: Spiral imaging has dramatically increased the diagnostic capabilities of Computed Tomography (CT) in the evaluation of small hepatocellular carcinomas (HCCs, O < 3 cm). We report our experience with multiple-phase CT of small HCC relative to both examination technique and lesion patterns. We compared the yield of biphasic (arterial-dominant + portal-dominant phases) and triphasic (arterial-dominant + portal-dominant + delayed phases) and also reviewed the literature for a meta-analysis of the techniques used.

MATERIAL AND METHODS: December 1996 to July 1998, forty-eight patients with small nodular HCCs were examined--98 nodules in all (range 1-9, mean 2 per patient). After baseline CT, a nonionic contrast agent (350 mgI/mL, 130-140 mL, 4 mL/s) was administered through a power injector and a 16-gauge needle. Biphasic volume images were acquired in 19 subjects (early-phase delay 24 s, venous-phase delay 75 s) and triphasic images in 29 (early-phase delay 24 s, venous-phase delay 60 s, delayed-phase delay 100 s). Retrospectively we assessed the number of nodules detected with each protocol in every phase, nodule conspicuity (graded I-IV) relative to surrounding parenchyma, and nodule patterns in the various phases. Nodule patterns were distinguished into homogeneous, peripheral, central and mixed hyperdensity, and homogeneous hypodensity.

RESULTS: Thirty-seven lesions were found in the patients examined with the biphasic technique: baseline images showed 35% of the nodules, arterial images 92%, portal images 76% and combined arterial and portal acquisitions 95%. Sixty-one lesions were found in the patients examined with the triphasic technique: baseline images showed 43% of the nodules, arterial images 93%, portal images 70%, and delayed images 77%; combined arterial and portal acquisitions detected 93% of the nodules, combined arterial and delayed images 95%, combined arterial and delayed images 80%. Finally, 95% of lesions were demonstrated when the three phases were combined. Overall conspicuity grades were I in 44% of cases, II in 28%, III in 18% and IV in 10% of cases at baseline scanning; I in 9%, II in 24%, III in 34% and IV in 33% in the arterial phase; I in 28%, II in 41%, III in 18% and IV in 13% in the portal phase; I in 23%, II in 30%, III in 26% and IV in 21% of cases in the delayed phase. At baseline, 10% of lesions were hyperdense (homogeneously and peripherally in 5% each); mixed density was seen in 8%, and hypodensity in 82%. In the arterial phase, 93% of lesions were hyperdense (homogeneously in 80%, peripherally in 10% and centrally in 3%); mixed density was seen in 5%, and hypodensity in 1%. In the portal phase, 4% of lesions were hyperdense (homogeneously in 1% and centrally in 3%); mixed density was seen in 11%, and hypodensity in 85%. In the delayed phase, the lesions appeared mixed in 11% of cases and hypodense in 89%.

CONCLUSIONS: Spiral CT scanning of small HCCs requires dedicated and meticulous technique. Multipassage assessment is mandatory, with 2 or 3 dynamic acquisitions of the whole liver. No major difference in nodule detection was demonstrated between these two options and thus the choice rests with the radiologist's preference. Early CT images proved best for lesion detection, followed by delayed, venous, and baseline studies; lesion recognition depends largely on nodular diameter. The same applies to conspicuity, which however depends on tumor volume less. Lesion patterns are quite typical and constant in all phases and are independent of lesion diameter.