Our prospective study was approved by the institutional review board and informed consent was obtained from all participant patients. In this study, a viable HCC around retained iodized oil was defined when local tumor progression was found around HCC with compact iodized oil accumulation by a prior TACE or when an incompletely treated viable HCC was found on follow-up CT or MR images following a TACE. From September 2008 to May 2009, patients with viable HCCs around the retained iodized oil by a prior TACE were enrolled, if they met the following inclusion criteria: 1) viable HCC around the retained iodized oil after a TACE was found at a follow-up liver CT and/or MRI, 2) in cases with a single viable HCC, the size of tumor was smaller than 5 cm in diameter, which was measured by the longest tumor diameter, including both viable HCCs and retained iodized oil based on CT or MR images, 3) in cases with multinodular HCCs (n ≤ 3), where each tumor had a longest diameter under 3 cm and at least one tumor was a viable HCC around retained iodized oil, 4) no portal vein invasion, 5) no extrahepatic metastases; 6) Child-Pugh class A or B; and 7) absence of severe coagulopathy (i.e., prothrombin activity < 40% or platelet count < 40000/mL).

Exclusion criteria were as follows: 1) the retained iodized oil around viable HCC was poorly visible on an anteroposterior and/or lateral projection image on a planning fluoroscopy for RFA; 2) in cases with multinodular HCCs, any new HCCs without retained iodized oil were poorly visible on planning US, and thus unsuitable for percutaneous US-guided RFA (even though the index tumor with retained iodized oil was visible on planning fluoroscopy); 3) percutaneous RFA was infeasible due to the high risk location of the collateral thermal injury or could result in incomplete ablation due to not having an adequate electrode path; 4) monoplane fluoroscopy was used due to the unavailability of biplane fluoroscopy at the time of the RFA procedure.

The diagnosis of a viable HCC around the retained iodized oil was based on the typical imaging features (arterial enhancement followed by delayed washout) on dynamic contrast-enhanced CT and/or MRI (11-13). Histopathologic confirmation was not obtained in any patient. The subphrenic location of the tumor was defined when the index tumor was either located within 1 cm from or abutting the diaphragm.

One abdominal radiologist performed both planning of the US fluoroscopy on the same day, to assess the feasibility of percutaneous RFA for all referred patients meeting the above inclusion criteria. The radiologist had 4 years of experience in performing percutaneous RFAs of liver tumors (more than 150 cases), at the starting point of this study. All planning US were performed with a multi-frequency 4C1 convex array probe (Acuson Sequoia 512, Siemens Medical Solutions, Mountain View, CA, USA). At the time of the planning US, the radiologist was aware that the intent of the RFA was to cover not only the viable HCC, but also the retained iodized oil; thus, lesion conspicuity was determined based on the US findings of the whole tumor. The radiologist searched thoroughly for the viable HCC with retained iodized oil, and determined whether the percutaneous US-guided RFA was feasible (14). The lesion conspicuity of the index tumor on planning US was classified as follows: visible group, conspicuous enough for percutaneous US-guided RFA; invisible group, inconspicuous and thus unsuitable for percutaneous US-guided RFA.

In addition, all patients with viable HCC around the retained iodized oil underwent a planning fluoroscopy with a flat-panel radiographic/fluoroscopic table system (SONIALVISION Safire 2; Shimadzu, Kyoto, Japan) by the same radiologist. The radiologist determined whether the retained iodized oil in the index tumor was radio-opaque enough to be seen on both anteroposterior and lateral projections, and thus could be targeted under the biplane fluoroscopy. Anteroposterior and lateral projection images were obtained with the patients in the supine and left lateral decubitus position, respectively. In terms of visibility of the index tumor on fluoroscopy, the HCCs were divided into two groups. When the retained iodized oil in the tumors was visible on both anteroposterior and lateral projection images, thus rendering the index tumor suitable for targeting under the guidance of biplane fluoroscopy, they were regarded as the fluoroscopically visible group. If the index tumors were invisible or poorly visible on anteroposterior, lateral or both images, thus not feasible for targeting the tumor, they were regarded as the fluoroscopically invisible group.

Percutaneous RFA was performed for the fluoroscopically visible group by the same radiologist in the interventional suite equipped with a biplane fluoroscopy/angiography instrument (Allura Xper FD 20/10, Philips Healthcare, Best, the Netherlands). The patients were placed in the supine position and were treated under local anesthesia or general anesthesia depending on tumor size and location. For the RFA procedure, a 200-W generator (Cool-tip; Valley lab, Boulder, CO, USA) and a 17-gauge cooled-tip electrode with a 3 cm exposed tip (Cool-tip; Valleylab, Boulder, CO, USA) or 18-gauge active-tip internally cooled electrode (Well-point RF Electrode; Taewoong Medical, Gimpo, Korea) were used. Artificial ascites were infused whenever they were required to improve the sonic window and decrease the degree of collateral thermal injury to the adjacent diaphragm or colon (15, 16). The intent of the RFA was to cover not only a viable HCC, but also the retained iodized oil where local tumor progression could subsequently occur (17). Targeting the index tumor was primarily guided by biplane fluoroscopy, using the retained iodized oil as an anatomic landmark. US guidance throughout the procedure was also needed to determine the skin entry site of the electrode, to avoid traversing dangerous structures such as the colon and the gallbladder, to estimate the ablation zone during the procedure and to avoid collateral thermal injury. Since the viable HCC itself was invisible on fluoroscopy, we placed an electrode based on CT or MR findings showing anatomic relationship between the viable HCC and the retained iodized oil. If viable HCC was visible on US, we used both US and biplane fluoroscopy as guiding modalities. However, when the viable tumor was invisible on US, biplane fluoroscopy was mainly used for the placement of an electrode during the ablation procedure, while US was used for monitoring the ablation process. Multiple overlapping ablations were facilitated by the retained iodized oil, which remained nearly unchanged after repeated cycles of ablation under biplane fluoroscopic guidance. Thus, with the multiple overlapping strategy, we could adopt a short ablation time, rather than the standard 12 minute-single ablation (i.e., 6 minutes × 2 overlapping ablations, rather than 12 minutes × single ablation), because we believe that the former could create a larger ablation zone than the latter (18). For hepatic dome lesions, the transthoracic approach (which can be complicated by the occurrence of a pneumothorax) could be avoided by the oblique approach from the lower intercostal space, under the guidance of biplane fluoroscopy.

If the index tumor was closely located to a critical structure, such as the central bile duct or large vessels, percutaneous ethanol injection (PEI) was combined with RFA simultaneously to avoid thermal injury and the heat-sink effect. Generally, absolute ethanol was injected to the parts of the tumor abutting or closely located to the critical structure and a RFA was applied to the other parts of the tumor located away from critical structure. First, a 21-gauge needle with six side holes (PEIT; Hakko, Tokyo, Japan) was introduced into the index tumor under the dual guidance of biplane fluoroscopy and US, and then the RF electrode was placed. The amount of ethanol injected into the tumor was always kept at less than the estimated volume (V): V (mL) = 4/3 π (γ + 0.5)³, where γ represents the radius of the tumor in centimeters. We injected absolute ethanol slowly into the deepest part of the tumor from the skin puncture site, and then into the superficial part of the tumor. The injection of ethanol was stopped when the echogenic zone was encompassing the parts of the tumor abutting or closely located to the critical structure and high resistance when ethanol injection was felt. Immediately after the PEI procedure, a RFA was performed. After the RFA procedure, the PEI needle was withdrawn slowly.

A CT scan was performed immediately after a RFA to evaluate the technical success and immediate post-RFA complication. Various multi-detector row CT scanners (Brilliance 40 [Philips Medical Systems, Cleveland, OH, USA], Lightspeed VCT [GE Medical Systems, Milwaukee, WI, USA], or Aquilion [Toshiba America Medical Systems, Tustin, CA, USA]) were used. A total of 120 mL of non-ionic contrast material (Ultravist 300, 300 mg/mL iopromide, Schering AG, Berlin, Germany) was administered intravenously, using an automatic injector at a rate of 3-4 mL/s. Triple phase CT Images from the dome of the diaphragm to the lower pole of the right kidney were obtained during a single breath-hold at 25-35, 60-70, and 180 seconds after initiating contrast material injection, representing the hepatic arterial, portal venous, and equilibrium phases, respectively. Axial images were reconstructed at a 5.0 mm slice thickness and reconstruction interval. Coronal and sagittal images were also reconstructed with both 3.0 mm slice thickness and reconstruction interval.

Technical success was defined as the presence of a non-enhancing area surrounding the tumor on CT images (19, 20). It was subdivided into two categories, primary and secondary technical success (Fig. 1). If only viable tumor around retained iodized oil was completely ablated on immediate follow up CT scan, it was defined as a primary technical success. When both viable tumor and retained iodized oil were completely ablated, it was defined as a secondary technical success. When primary technical success was not achieved, a second RFA session was performed within 24 hours. However, when secondary technical success was not achieved, we did not perform a second RFA session, since there was no evidence of a viable HCC based on CT images. Instead of an additional RFA session, a close follow-up was adopted for these patients.

After a post-RFA CT scan, the patients were followed up 1 month after the RFA and every 3 months thereafter, using liver function tests, serum α-fetoprotein, and dynamic liver CT. The primary end point was the rate of technical success and major complication. Major complications were defined as any event leading to substantial morbidity and disability, increasing the level of care, and resulting in substantially lengthening a hospital stay, requiring a blood transfusion, or an interventional drainage procedure (21). All others were considered to be minor complications. The secondary end point was the rate of local tumor progression. The rate of local tumor progression and distant metastasis (both intra- and extrahepatic metastasis) at follow-up CT were assessed. Local tumor progression was defined as a new enhancing lesion within or adjacent to the previously completely ablated site on dynamic liver CT/MR imaging (22). Distant metastasis was defined as a new HCC in the liver, located distant from the index tumor or in extrahepatic regions.

Statistical analyses were performed with the MedCalc software version 9.3.9.0 (MedCalc Software, Mariakerke, Belgium). The fluoroscopically visible group was compared with the fluoroscopically invisible group in terms of the size of retained iodized oil, location of tumor (subphrenic or not), and lesion conspicuity on planning US using the Mann-Whitney test or chi-square test. P < 0.05 was considered to indicate a statistically significant difference.

During the study period, 40 consecutive patients met the inclusion criteria and were referred for planning US and planning fluoroscopy for percutaneous RFA (Fig. 2). Among the 37 patients with a single viable HCC around the retained iodized oil, 25 patients had fluoroscopically visible tumors, whereas the other 12 patients had fluoroscopically invisible tumors. Of the 25 patients with fluoroscopically visible tumors, four patients who had tumors in high risk locations (n = 2) or were treated under the guidance of monoplane fluoroscopy and US (n = 2) were excluded. The remaining 21 tumors were ablated under the guidance of biplane fluoroscopy assisted by US guidance. On the other hand, 3 patients with multiple HCCs were excluded because the viable HCC around the retained iodized oil was invisible on fluoroscopy (n = 1) or new HCCs were invisible on US (n = 2). Finally, a total of 21 patients with 21 viable HCCs with retained iodized oil were treated under the guidance of biplane fluoroscopy assisted by US guidance, and were included in the final analysis in this study (Figs. 2, 3). Baseline characteristics of 21 patients are summarized in Table 1. Ten patients had residual untreated viable HCCs after a TACE, while the other 11 patients had local tumor progression around HCC with compact iodized oil accumulation on follow-up CT images. The number of previous TACE procedures on viable HCCs with retained iodized oil before a RFA was 3.4 ± 2.4 (range: 1-10). The time interval between the last TACE and the ablation procedure was 9.4 ± 8.7 months, ranging from 1.0 to 29.7 months. The size of the retained iodized oil and viable portion of the HCC ranged from 0.5 to 4.2 cm (mean ± standard deviation [SD]: 2.8 ± 1.1 cm) and 0.8 to 4.5 cm (mean ± SD: 1.9 ± 1.0 cm), respectively. The total size of the tumor, including both the viable HCC and retained iodized oil, ranged from 1.4 to 5.0 cm (mean ± SD: 3.2 ± 1.0 cm). Twelve of 21 (57.1%) HCCs were located in the subphrenic region and the other 9 (42.9%) were located elsewhere.

Artificial ascites were introduced in 16 patients to minimize thermal injury to the diaphragm or colon. The number of overlapping ablations was 4.3 ± 2.3 per HCC nodule (range: 1-8). The total ablation time was 22.3 ± 13.1 minutes (range: 7-56 minutes) (Table 2).

On the planning fluoroscopy for a RFA, among the 42 viable HCCs with retained iodized oil in the 40 patients, 27 (64.3%) tumors were fluoroscopically visible, whereas 15 (35.7%) were fluoroscopically invisible (Table 3). Among the 27 fluoroscopically visible tumors, 6 were not ablated under biplane fluoroscopy-guidance because of the following reasons: 1) RFA was not performed since the viable HCCs were located in high risk regions (n = 2); 2) RFA was performed under monoplane and US guidance (n = 2); and 3) in cases with multi-nodular HCCs, although the viable tumors with retained iodized oil were visible on fluoroscopy, while other new HCCs were invisible on US (n = 2) (Fig. 2).

For all 42 viable tumors around the retained iodized oil, which were evaluated by planning fluoroscopy and US, the size (2.7 ± 1.1 cm) of the retained iodized oil in the fluoroscopically visible tumors was significantly larger than that (1.0 ± 0.6 cm) of fluoroscopically invisible tumors (p < 0.001). However, there was no significant difference in the location of tumor (subphrenic or not) between the two groups (p = 0.818). For the lesion conspicuity on US, there was also no significant difference between the two groups (p = 0.079) (Table 3).

Among the 21 viable HCCs around the retained iodized oil which were treated under the guidance of biplane fluoroscopy and US, 12 (57.1%) tumors were located in the subphrenic region and the other 9 (42.9%) in the non-subphrenic region. Although 15 (71.4%) tumors were conspicuous enough for US-guided RFA, the other 6 (28.6%) tumors were invisible on US.

On immediate post-RFA CT images, the long and short diameters of the ablation zone were 4.9 ± 1.1 cm and 3.5 ± 1.0 cm, respectively. Primary technical success was achieved in all 21 HCCs. Secondary technical success was achieved in only 17 of 21 (81.0%) HCCs. However, a second RFA session was not performed in any of the 4 patients, because there was no evidence of viable HCCs on CT images. Both the sizes of the retained iodized oil (median, 2.5 cm; range, 0.5-4.2 cm vs. median, 3.8 cm; range, 2.8-4.0 cm; p = 0.066) and total tumor (median, 3.0 cm; range, 1.4-4.5 cm vs. median, 3.8 cm; range, 3.6-5.0 cm; p = 0.140), were slightly smaller in the secondary technical success group than the secondary technical failure group. However, the differences were not statistically significant. Major complications were not observed in any patient.

During follow-up period (20.3 ± 7.4 months; range, 6.5-29.9 months; and median, 22.6 months), the rate of local tumor progression was 9.5% (2/21). Because these two patients with local tumor progression had synchronous distant intrahepatic metastases, repeated TACE was performed for further management. Distant intrahepatic metastasis developed frequently in 76.2% (16/21) of patients.