DISCUSSION

RF ablation is a thermal ablation procedure and produces localized and controlled tumor destruction. Alternating electric current conducted by an electrode produces local ionic agitation and subsequent friction heat, resulting in coagulation necrosis. It is usually performed in primary or secondary hepatic malignancies of 5cm or smaller, four or fewer, and without extra-hepatic tumor (10). Transcatheter arterial chemoembolization produces ischemic necrosis of the tumor by the selective embolization of the hepatic arterial supply. Transcatheter arterial chemoembolization has the capability of treating the entire liver and can be performed irrespective of the size and the location of the hepatocellular carcinoma. So transcatheter arterial chemoembolization can be performed before the RF ablation of the hepatocellular carcinoma to reduce tumor perfusion and heat loss (11, 12), and the transcatheter arterial chemoembolization can be performed after the RF ablation when the local tumor progression or remote intrahepatic recurrence has occurred. The procedure of RF ablation resulted in various alterations in the angiographic findings, which may affect the procedure of trancatheter arterial chemoembolization.

The arterio-portal shunt may develop spontaneously in a cirrhotic liver due to the structural distortions, due to the hepatocellular carcinoma, or iatrogenically along the path of a biopsy needle after a liver biopsy procedure. The frequency of the arterio-portal shunt after puncturing the liver with a needle depended on the interval between the needle puncture and the hepatic arteriography, as high as 50% within the first week and less than 10% after the first week (18). The arterio-portal shunt tended to close spontaneously in time. The arterio-portal shunt also occurred after the RF ablation procedure. The arterio-portal shunt

or by the thermal injury of the RF ablation. However, the staining of the hepatocellular carcinoma at the margin or adjacent to the ablation zone can be hidden within the staining of the arterio-portal shunt, and localization for super-selective transcatheter arterial chemoembolization will be difficult.

Furthermore, the iodized oil/doxorubicin emulsion will bypass the arterio-portal shunt and selective uptake by the hepatocellular carcinoma will be prohibited (Fig 1).

The periablational enhancement seen in the contrast enhanced CT correlated with a rim constituting of a mixture of viable hepatic parenchyma, necrosis, and hemorrhage (13). The enhancement was hyper-attenuating in the arterial phase, and differentiation from the locally progressed hepatocellular carcinoma was difficult (17). The frequency of the periablational enhancement also depended upon the interval from the RF ablation to the CT, and persisted as long as six months after the RF ablation (17). Differentiating the locally progressed hepatocellular carcinoma from the surrounding periablational enhancement on hepatic arteriography was also difficult (Fig 2) and precise localization for the super-selective transcatheter arterial chemoembolization was not possible. The periablational enhancement also persisted as long as 196 days after RF ablation in the hepatic arteriograms.

The caliber of an artery will vary with the arterial demand of the tissue the artery is supplying. When the size of the locally progressed hepatocellular carcinoma of the post-RF ablation hepatic arteriography has varied compared to the index tumor, the caliber of the feeding artery will also vary (Fig 3). The advances of the quality of the microcatheters manufactured recently have made super-selective transcatheter arterial chemoembolization feasible, and most feeding arteries to the hepatocellular carcinomas can be easily super-selected. But the decreased caliber of a feeding artery to a hepatocellular carcinoma will increase the difficulty of super-selective transcatheter arterial chemoembolization, especially when fine collaterals are feeding the hepatocellular carcinoma.

When RF ablation of a hepatic malignancy is performed, some amount of the normal liver parenchyma adjacent to the tumor is also ablated to acquire the

ablative margin of 0.5 to 1 cm. There is a high possibility that a subsegmental artery adjacent to the tumor will be supplying the normal liver parenchyma adjacent to the tumor, especially at subcapsular region. Occlusion of a subsegmental artery adjacent to a subcapsular hepatocellular carcinoma after RF ablation may have resulted from direct thermal injury by the RF ablation, or the diminished demand by the liver parenchyma adjacent to the hepatocellular carcinoma. Occlusion of a subsegmental artery in a deep region of the liver parenchyma was seen in one case. Intra-hepatic collaterals were supplying the liver parenchyma distal to the occluded artery (Fig 4). When a recurrent hepatocellular carcinoma has occurred distal to an occluded artery deep in the liver parenchyma, it will also be supplied by the intra-hepatic collaterals, which will increase the difficulty of super-selective transcatheter arterial chemoembolization.

There are some limitations in the results of our study. The study was retrospectively designed and is descriptive. Hepatocellular carcinomas were pathologically diagnosed in only twelve patients. Locally progressed or remote intrahepatic recurrent hepatocellular carcinomas after RF ablation were radiologically diagnosed, but not pathologically diagnosed. The post-RF ablation hepatic arteriography was performed only when recurrent hepatocellular carcinoma was suspected on follow up liver CT scans resulting in a selection bias and the number of hepatic arteriography within 3 months from RF ablation is small. The arterio-portal shunt may have been induced by the RF electrode, but not by the thermal injury of the RF ablation. And the changes of the portal vein and the hepatic vein could not be evaluated.