Original Research
Vascular and Interventional Radiology
September 2010

The Minimal Ablative Margin of Radiofrequency Ablation of Hepatocellular Carcinoma (> 2 and < 5 cm) Needed to Prevent Local Tumor Progression: 3D Quantitative Assessment Using CT Image Fusion

Abstract

OBJECTIVE. The aim of this study was to elucidate the minimal ablative margin for percutaneous radiofrequency ablation (RFA) of hepatocellular carcinoma (HCC) (> 2 and < 5 cm) needed to prevent local tumor progression using CT image fusion and a 3D quantitative method.
MATERIALS AND METHODS. From April 2005 to March 2007, we performed percutaneous RFA for the treatment of 382 HCCs larger than 2 cm and smaller than 5 cm. A total of 110 tumors in 103 patients (77 men and 26 women; mean age, 59.7 years) that were previously untreated and were monitored for at least 1 year were retrospectively enrolled. A 5-mm safety margin was attempted in all cases, and a CT finding of complete replacement of the index tumor by RFA zone was defined as technical success. We constructed fusion images of CT images obtained before and after RFA and performed radial multiplanar reformation with the rotation axis at the center of the tumor to analyze the ablative margin quantitatively. Risk factors for local tumor progression (the thinnest ablative margin, tumor size, and the effect of hepatic vessels) were assessed by multivariate analysis.
RESULTS. Patients underwent follow-up for 12.9–46.6 months (median, 28.1 months). The tumors were 2.1–4.8 cm (mean ± SD, 2.7 ± 0.6 cm) in diameter. The thinnest ablative margins ranged from 0 to 6 mm (1.0 ± 1.4 mm). A 5-mm safety margin was achieved in only 2.7% (3/110) of cases. In 47.3% (52/110) of cases, vessel-induced indentation of the ablation zone contributed to the thinnest ablative margins. Local tumor progression was detected in 27.3% (30/110) of cases. Concordance between local tumor progression and the thinnest margin was observed in 83.3% (25/30) of cases. The incidence of concordant local tumor progression was 22.7% (25/110), 18.9% (10/53), 5.9% (2/34), and 0% (0/15) in tumors with the thinnest ablative margin of ≥ 0, ≥ 1, ≥ 2, and ≥ 3 mm, respectively. An insufficient ablative margin was the sole significant factor associated with local tumor progression.
CONCLUSION. When the thickness of the ablative margin is evaluated by CT image fusion, a margin of 3 mm or more appears to be associated with a lower rate of local tumor progression after percutaneous RFA of HCC.

Introduction

For radiofrequency ablation (RFA) of hepatocellular carcinoma (HCC), there are two types of intrahepatic recurrence: local tumor progression and intrahepatic distant recurrence, according to the location of the tumor. By definition [1], local tumor progression refers to a recurrence either abutting or within the previous RFA zone; the reported incidence of local tumor progression has ranged from 0.9% to 31.3% [29]. An intrahepatic distant recurrence is known to be associated with changes in the serum α-fetoprotein level, the multifocality of the HCC, or hepatitis C virus infection [5, 10, 11]. The known risk factors for local tumor progression include large tumor size [2, 3, 5, 6], an insufficient ablative margin [5, 9], blood vessels close to the tumor [5, 9], poor histologic grading [4], and viable tumor cells adherent to the electrode used for RFA [12]. These factors correlate with the ablation results, tumor characteristics, and the local environment surrounding the tumor rather than the systemic conditions.
Several investigators have studied the relationship between the ablative margin and recurrence of HCC [5, 9, 11]. However, in all previous studies, the ablative margin was assessed by comparing axial CT images obtained before and after RFA in a side-by-side manner. This approach is limited by the ability to assess the ablative margin exactly, especially in cases with a vertical or oblique dimension. In addition, even though a specific cut-off value (5 or 10 mm) for the ablative margin has been adopted and the significance of the margins for predicting local tumor progression have been evaluated, the specific values used were arbitrary or, at best, based on the results of several studies that evaluated the safety margin of surgical resection associated with HCC [1315]. To our knowledge, no prior study has evaluated the safety margin for RFA of HCC in an objective and quantitative manner. To overcome the drawbacks of previous studies, we used a CT workstation that was capable of fusing CT image sets obtained before and after RFA. In addition, radial multiplanar reformation of the fused images was performed with the axis of rotation at the center of the tumor to view the actual gap between the treated HCC and the ablation zone in virtually every direction of the index tumor.
The aims of this study were to validate the feasibility of a 3D quantitative method using CT image fusion and radial multiplanar reformation for analyzing the ablative margin after RFA, to elucidate the minimal ablative margin required for preventing local tumor progression after RFA of HCC measuring larger than 2 cm and smaller than 5 cm in diameter, and to understand the effects of hepatic vessels on the ablative margins in an objective and quantitative way.

Materials and Methods

Patients and Tumors

Approval was obtained from the institutional review board, and informed consent for the percutaneous RFA procedures and for the use of data for research was obtained from all patients. From April 2005 to March 2007, we performed percutaneous RFA under ultrasound guidance for the treatment of 382 HCC tumors larger than 2 cm and smaller than 5 cm in diameter in 320 patients (235 men and 85 women; age range, 35–88 years; mean age, 58.2 years) with the use of an internally cooled electrode system (Cool-tip RF Ablation System, Covidien). Among the treated tumors, 146 were excluded from the study for the following reasons: a follow-up period of less than 1 year (n = 102), partial treatment by previous chemoembolization (partial iodized oil retention) (n = 20), technical failure of RFA (n = 18), and the presence of disseminated recurrent tumors that abutted the RFA zone and were deemed to be intrahepatic distant recurrence because of their distribution and sudden simultaneous occurrence (n = 6).
Among the remaining HCC tumors, 110 tumors in 103 patients (77 men and 26 women; age range, 39–83 years; mean age, 59.7 years) were finally chosen for this study on the basis of the following selection criteria: availability of baseline CT performed within 1 month before the procedure and availability of immediate CT study performed within 24 hours. Cases were excluded if there was a significant discrepancy in the liver configuration due to respiration that hindered correct image registration for fusion and if the quality of the baseline CT image that was obtained at an outside hospital was poor as a result of the use of large slice thickness or a DICOM-incompatible imaging format. Among these 110 HCC tumors, 26 (23.6%) were confirmed by histopathologic analysis, and the remaining 84 tumors (76.4%) were considered HCC on the basis of the clinical criteria from the American Association for the Study of Liver Diseases [16].
Percutaneous RFA was performed when all of the following criteria were met: the patient had a single HCC 5 cm or less in maximal diameter, the patient had a multinodular HCC (≤ 3 in number) with each tumor measuring up to 3 cm in maximal diameter, the patient had liver disease of Child-Pugh class A or B, tumors could be visualized by ultrasound and were accessible with a percutaneous route, the prothrombin time ratio was greater than 50% (prothrombin time with international normalized ratio < 1.7), and the platelet count was greater than 50,000 cells/mm3 (50 cells × 109/L) [2]. If the size and number of the tumors were within the defined criteria and if the index tumor was not clearly visible even after contrast-enhanced ultrasound or the introduction of artificial ascites, chemoembolization was first performed, and then percutaneous RFA was performed under fluoroscopy or cone-beam CT guidance or both in a single session. If the size and number of index tumors did not conform to the criteria, chemoembolization was first performed and percutaneous RFA was added when necessary. However, the final decision about the procedure to be used was made by consensus of an interdisciplinary tumor board. The baseline characteristics of the study population are summarized in Table 1.
TABLE 1: Baseline Features of Study Population
Clinical CharacteristicsValue (n = 103 Patients)
Ratio of men to women77:26
Age range (y)39–83
Liver disease 
    Liver cirrhosis89
    Chronic hepatitis9
    Healthy carrier4
    None1
Cause of liver disease 
    Hepatitis B virus75
    Hepatitis C virus25
    Non-B and non-C hepatitis virus2
    Alcohol0
    Cryptogenic0
    None1
Child-Pugh classification 
    Class A79
    Class B24
Serum α-fetoprotein level, ng/mL, mean ± SD (range)105.0 ± 226.5 (0–1,288)a
    < 2041
    20–20027
    > 20011
Tumors (n = 110) 
    Size, cm, mean ± SD (range)2.7 ± 0.6 (2.1–4.8)
    Diagnosis of hepatocellular carcinoma 
        Histopathology26
        Clinical criteria84
    History of hepatocellular carcinoma treatment 
        No52
        Yes51
    Location 
        Subcapsular59
        Nonsubcapsular44
RFA procedure (n = 103) 
    Electrode used 
        Single type (3 cm)76
        Cluster type22
        Single plus cluster5
    Ablation time, min, mean ± SD (range)15.0 ± 4.8 (6–30)
    No. of overlapping ablations25
    No. of times overlapping, range2–4
    Use of artificial ascites
12
Note—Except where noted, data are no. of patients. RFA = radiofrequency ablation.
a
Data were available for 79 patients.

RFA Procedure

All procedures were performed under ultrasound guidance by one of five operators who had at least 3 years of experience with this procedure at the time of the earliest RFA procedure. Details of the technique used have been described elsewhere [17]. We used an internally cooled electrode system for all procedures. Two types of electrodes were used with a 200-W radiofrequency generator: a single 17-gauge straight electrode with a 3-cm active tip, or a cluster type electrode with a 2.5-cm active tip. Our strategy for complete necrosis of the tumor was to ablate at least 0.5 cm of the normal hepatic parenchyma surrounding the tumor as a safety margin. Therefore, for tumors larger than 2.5 cm in diameter, we used a multiple overlapping ablation technique or a cluster type electrode. The decisions with regard to electrode type and overlapping ablation technique were discussed at an intradisciplinary meeting before treatment; however, the final decision was left to the discretion of the operator after considering the size and configuration of the tumor. There was no limitation based on the proximity of the index tumor to the adjacent hepatic vessels. We used the artificial ascites technique when the anticipated RFA zone was in contact with a critical organ, such as the hepatic flexure of the colon, to prevent collateral thermal injury or when the index tumor was located in the hepatic dome area to improve visibility (13.6% [14/103]) [18].

Follow-Up After Treatment

All patients underwent follow-up with contrast-enhanced three-phase CT examinations. For early evaluation of the therapeutic response or possible complications, CT examination was performed within 24 hours if the condition of the patient allowed. A follow-up study was performed 1 month later. If the RFA was considered to be technically effective, a follow-up CT scan was repeated every 3 months after this visit.
Contrast-enhanced CT examinations were performed with one of two MDCT scanners (Brilliance 40, Philips Healthcare; or Aquilion 64, Toshiba Medical Systems Corporation). We used 120 mL of nonionic contrast material (iopromide; Ultravist 300, Bayer Schering Pharma) that was administered IV with an automatic injector at a rate of 3 mL/s. The images were obtained at 20–35, 70, and 180 seconds after the contrast material was administered by the bolus-tracking technique, at the hepatic arterial, portal venous, and equilibrium phases. The parameters for CT were a 1.25-mm slice thickness and a 1.25-mm reconstruction interval.
Local therapeutic efficacy was assessed in terms of technique effectiveness and local tumor progression. The incidence and estimated cumulative rates of local tumor progression at each year of follow-up were also calculated.

Definitions of Terminology

Definitions are essentially based on the standardization of terminology and reporting criteria published by the International Working Group on Image-Guided Tumor Ablation [1]. Technical success was defined when the tumor was treated according to protocol and was completely replaced by the RFA zone at immediate follow-up CT scan. Otherwise, the treatment was defined as technical failure. Achievement of technique effectiveness was defined when complete ablation of macroscopic tumor was evidenced by 1-month follow-up CT scan. Local tumor progression was diagnosed when a follow-up CT scan revealed interval development or growth of the tumor or new iodized oil density retained along the margin of ablation zone by chemoembolization, which was performed for intrahepatic distant recurrence where the RFA had been considered to be technically effective. When both locations of the thinnest ablative margin and local tumor progression were within ± 1 o'clock (explained later), this was defined as concordant local tumor progression. The reason why we regarded ± 1 o'clock difference as concordant was that atrophy and deformity of the ablation zone increased with time. Otherwise, local tumor progression was defined as discordant.

Image Processing for Quantitative Analysis of Ablative Margins

We used a CT workstation (Virtual Place Advance Plus version 2.03, Aze Corporation) that was capable of performing image fusion of CT studies before and after the RFA procedure. The CT data were acquired from the archive of the CT workstation, which had a 1.25-mm interval. One radiologist performed the CT image fusion for all patients using the manual segmentation registration method. Among the three phases of CT performed before RFA, the one phase where the tumor size was the greatest was selected, and the same phase of CT performed after RFA was fused to it. Image registration was performed by aligning two overlaid CT studies with six parameters, translation and rotation in three reformed planes, to maximize image similarity around the tumor and the ablation zone. The hepatic contour and the hepatic artery–portal vein complex or hepatic vein near the tumor were used as landmarks for fine adjustments. When image registration was satisfactory, the fused images were reconstructed. Using the reconstructed fusion images, radial multiplanar reformation was performed with the axis of rotation at the center of the tumor on the axial plane. The parameters of radial multiplanar reformation were as follows: slab thickness, 1 mm; slice interval, 5°; and reconstruction mode, volume rendering (Figs. 1A, 1B, 1C, 1D, and 1E). All radiologists who participated in this study had at least 5 years of experience in interpretation of liver CT scans. If multiple tumors were present in the same patient, the aforementioned image processing from registration to radial multiplanar reformation was repeated for each tumor.

Quantitative Analysis of the Ablative Margins

To describe the direction of the thinnest ablative margin and local tumor progression, we devised a 12 o'clock–6 o'clock coordinate system, as shown in Figure 2. The radial multiplanar reformation images were assessed by two radiologists who were blinded to the locations of the local tumor progression, by consensus. The radiologists sought the thinnest ablative margin and recorded its location. When the ablation zone abutted the hepatic capsule, it was recorded, but it was ignored for the evaluation of the thinnest margin. The effect of a hepatic vessel (hepatic artery–portal vein complex or hepatic vein) on the formation of the thinnest ablative margin (i.e., the so-called heat-sink effect) was also assessed. The heat-sink effect was considered present when the ablation zone was indented by the hepatic vessels, and this indentation formed the thinnest ablative margin. An additional two radiologists who were blinded to the results of the ablative margin analysis reviewed all of the follow-up CT studies of the patients to evaluate the date and location of local tumor progression with the same radial multiplanar reformation and coordinate system by consensus. If the two evaluators could not reach consensus, a third radiologist made the decision. Finally, we evaluated the relationships between the location of the thinnest ablative margin and local tumor progression, whether or not they were concordant. Considering only concordant local tumor progression, the minimal ablative margin that prevented occurrence of local tumor progression was determined by assessing the incidence of local tumor progression at each thickness per millimeter.
Fig. 1A 43-year-old man with hepatocellular carcinoma (HCC) who was treated with percutaneous radiofrequency ablation (RFA). Single nodular HCC (arrow) measuring 2.8 cm is noted in center of right hepatic lobe on hepatic arterial phase of CT scan.
Fig. 1B 43-year-old man with hepatocellular carcinoma (HCC) who was treated with percutaneous radiofrequency ablation (RFA). Hepatic arterial phase of immediate follow-up CT scan after percutaneous RFA using cluster type internally cooled electrode shows complete ablation of HCC without unablated residual tumor. Thin enhancing line surrounding ablation zone represents transient hyperemia.
Fig. 1C 43-year-old man with hepatocellular carcinoma (HCC) who was treated with percutaneous radiofrequency ablation (RFA). Image registration of CT images obtained before and after RFA is performed using three rotation and three translation parameters. Overlay of CT images obtained before RFA (orange-color scale) and after RFA (gray scale) is shown.
Fig. 1D 43-year-old man with hepatocellular carcinoma (HCC) who was treated with percutaneous radiofrequency ablation (RFA). Radial multiplanar reformation images performed on axial fused image with its axis of rotation at center of tumor show that margins of index tumor and ablation zone abut each other in 5–6 o'clock (axial plane) to 4–5 o'clock (vertical plane) direction; thus, ablative margin is 0 mm (arrows).
Fig. 1E 43-year-old man with hepatocellular carcinoma (HCC) who was treated with percutaneous radiofrequency ablation (RFA). Unenhanced CT scan obtained 23 months after transcatheter arterial chemoembolization shows small high-attenuating nodule representing iodized oil retention of local tumor progression (arrow) at ablative margin of 4–5 o'clock to 4–5 o'clock direction, which could be regarded as concordant to thinnest ablative margin after considering deformity of ablation zone.

Qualitative Analysis of the Ablative Margins

To compare with the diagnostic performance of quantitative method, we performed qualitative side-by-side comparison of CT scans obtained before and after RFA with the use of axial CT images only. Two radiologists who were blind to the results of quantitative analysis assessed in consensus whether an ablative margin ≥ 5 mm was achieved. For this analysis, the adjacent hepatic vessels or the hepatic capsule were used to facilitate comparison.

Statistical Analysis

Cumulative local tumor progression rates were estimated using a survival curve constructed by the Kaplan-Meier method. The thicknesses of the thinnest ablative margin of the group with or without concordant local tumor progression were compared (Student's t test). Comparison between qualitative and quantitative methods was made by chi-square test. Multivariate analysis using Cox's regression hazard model was performed to determine which variable was independently significant for occurrence of concordant local tumor progression among the thinnest ablative margin, tumor size, and the effect of hepatic vessels. A p value < 0.05 was considered significant. Data were analyzed using commercially available statistics software (SPSS, version 9.0, SPSS) for Microsoft Windows.
Fig. 2 Coordinate system used to describe direction of local tumor progression from center of tumor. System consists of two numerals in which first one indicates direction on axial plane from 0 to 11 o'clock and second one indicates direction on vertical plane from 0 to 6 o'clock. Each direction line can be drawn on surface of tumor similar to longitude and latitude lines on terrestrial globe. Intersecting points made by each type of line created 62 coordinates, including top and bottom.

Results

Local Therapeutic Efficacy

The period of CT follow-up ranged from 12.9 to 46.6 months (median, 28.1). The technical effectiveness rate was 100.0% (110/110). The frequency of local tumor progression was 27.3% (30/110), which occurred at 3.3–40.5 months (mean, 37.7; standard error, 1.6 months; 95% CI, 34.6–40.8 months). The cumulative local tumor progression rates were estimated as 12.7% at 1 year, 26.4% at 2 years, 27.9% at 3 years, and 37.8% at 4 years (Fig. 3).

Quantitative Analysis of Ablative Margins and Local Tumor Progression

To perform image fusion, the hepatic arterial phase was chosen for 78 tumors, the portal venous phase was chosen for five tumors, and the equilibrium phase was chosen for 27 tumors. The thinnest ablative margin in 110 tumors ranged from 0 to 6 mm (mean ± SD, 1.03 ± 1.37 mm). The margin was 0 mm in 57 (51.8%) of the 110 tumors and 1 mm or thicker in 53 tumors (48.2%) (1 mm, n = 19; 2 mm, n = 19; 3 mm, n = 9; 4 mm, n = 3; 5 mm, n = 1; 6 mm, n = 2). In four cases, there were two equal foci of the thinnest margins (0 mm, n = 2; 1 mm, n = 1; 4 mm, n = 1). Qualitative analysis using side-by-side comparison showed that 34.5% (38/110) of ablations had established at least 5-mm ablative margins, which was significantly higher than the results as determined by quantitative analysis (2.7% [3/110]) (p < 0.0001).
Twenty-five (83.3%) of 30 local tumor progressions occurred at the point of the thinnest ablative margin, and the local tumor progressions were considered as concordant. The thickness of the thinnest ablative margin in the tumor group with concordant local tumor progression (0.48 ± 0.65 mm; range, 0–2 mm) was significantly smaller than in the tumor group without concordant local tumor progression (1.19 ± 1.48 mm; range, 0–6 mm; p = 0.001).
Fig. 3 Cumulative rates of local tumor progression after percutaneous radiofrequency ablation of hepatocellular carcinomas larger than 2 cm were estimated as 12.7% at 1 year, 26.4% at 2 years, 27.9% at 3 years, and 37.8% at 4 years by use of Kaplan-Meier method. Cross marks (+) represent censored data.
Considering only concordant local tumor progression, the frequencies of local tumor progression at each thickness of the thinnest ablative margin were 22.7% (25/110), 18.9% (10/53), 5.9% (2/34), and 0% (0/15) in tumors with the thinnest ablative margins of ≥ 0 mm, ≥ 1 mm, ≥ 2 mm, and ≥ 3 mm, respectively. If the study group is divided according to the thinnest ablative margin (< 3 and ≥ 3 mm), the frequencies of concordant local tumor progression were 26.3% (25/95) (follow-up period, 12.9–48.1 months; median, 28.6 months) and 0% (0/15) (follow-up period, 12.0–32.5 months; median, 24.8 months), respectively.

Effect of Hepatic Vessels on the Ablative Margin

In 47.3% (52/110) of the cases, formation of the thinnest ablative margin was attributed to hepatic vessel–induced indentation of the ablation zone (Figs. 4A, 4B, 4C, and 4D). The hepatic artery–portal vein complex and hepatic vein were responsible for 94.2% (49/52) and 5.8% (3/52), respectively, of cases of indentation. Another 52.7% (58/110) of the cases had no specific cause for the thinnest ablative margin other than the geographic relationship between the tumor and the ablation zone induced by a large tumor size, insufficient volume of the ablation zone, or inappropriate centering of the electrode to the tumor. Among 95 ablations that failed to establish at least a 3-mm ablative margin, 48 cases (50.5%) were affected by hepatic vessels (hepatic artery–portal vein complex, n = 45; hepatic vein, n = 3), and the remainder of cases had no specific cause identified.
Fig. 4A 56-year-old woman with hepatocellular carcinoma who was treated with percutaneous radiofrequency ablation (RFA). Each figure represents radial reformation images of fused CT scans obtained before and after RFA with 45° interval. Index tumor (asterisk) is completely replaced by RFA zone (arrowheads). However, tumor margin contacts with ablative margin as a result of effect of adjacent hepatic artery–portal vein complex (arrows, B), which means that thinnest ablative margin is 0 mm in this case.
Fig. 4B 56-year-old woman with hepatocellular carcinoma who was treated with percutaneous radiofrequency ablation (RFA). Each figure represents radial reformation images of fused CT scans obtained before and after RFA with 45° interval. Index tumor (asterisk) is completely replaced by RFA zone (arrowheads). However, tumor margin contacts with ablative margin as a result of effect of adjacent hepatic artery–portal vein complex (arrows, B), which means that thinnest ablative margin is 0 mm in this case.
Fig. 4C 56-year-old woman with hepatocellular carcinoma who was treated with percutaneous radiofrequency ablation (RFA). Each figure represents radial reformation images of fused CT scans obtained before and after RFA with 45° interval. Index tumor (asterisk) is completely replaced by RFA zone (arrowheads). However, tumor margin contacts with ablative margin as a result of effect of adjacent hepatic artery–portal vein complex (arrows, B), which means that thinnest ablative margin is 0 mm in this case.
Fig. 4D 56-year-old woman with hepatocellular carcinoma who was treated with percutaneous radiofrequency ablation (RFA). Each figure represents radial reformation images of fused CT scans obtained before and after RFA with 45° interval. Index tumor (asterisk) is completely replaced by RFA zone (arrowheads). However, tumor margin contacts with ablative margin as a result of effect of adjacent hepatic artery–portal vein complex (arrows, B), which means that thinnest ablative margin is 0 mm in this case.

Multivariate Analysis of Concordant Local Tumor Progression

Among the factors that influence local tumor progression (i.e., thinnest ablative margin, tumor size, and the effect of hepatic vessel on the ablation zone), the thinnest ablative margin was determined to be the sole factor that was independently significant for the occurrence of concordant local tumor progression according to the multivariate analysis (p = 0.042; B = –0.443; risk ratio = 0.642 [95% CI, 0.413–0.997]). The other two factors were not statistically significant (tumor size, p = 0.450; the effect of hepatic vessels, p = 0.585).

Discussion

Throughout the study period, the consistent strategy used for complete necrosis of tumors was to ablate at least 5 mm of the normal hepatic parenchyma surrounding the tumor as a safety margin. Setting 5 mm as the criterion was not decided on a scientific basis but rather according to the results of previous studies [5, 9] and according to the generally accepted surgical resection margins for HCC [1315] and the feasibility of RFA. Nevertheless, 97.3% (107/110) of cases failed to achieve a 5-mm margin. The results of this study showed that a 5-mm ablative margin in all surrounding directions of tumors larger than 2 cm was practically difficult on the basis of the analysis using the CT image fusion method. In many cases, failure was attributed to the effect of hepatic vessels. In other cases, we could not even detect the insufficiency of the ablative margin in the vertical or oblique direction by side-by-side comparisons.
If we divide the study group according to the thinnest ablative margin (≥ 3 or < 3 mm), the frequency of concordant local tumor progression was 0% and 26.3%, respectively. Thus, on the basis of CT image fusion analysis, a 3-mm ablative margin may be regarded as a minimum requirement to prevent local tumor progression after RFA of HCC. However, we do not recommend a 3-mm margin as a goal of treatment because of the limitations of this study, which will be discussed later in the article. Our 3-mm criterion may be used for determining follow-up rather than the goal of ablation—that is, if we ablated the entire visible tumor but failed to establish a 3-mm ablative margin, a closer follow-up schedule would be recommended.
The measurements of ablative margin thickness using only axial image sets compared with the 3D method can be different. For example, in the oblique direction, the thickness of the ablative margin evaluated with only axial images might exaggerate the real thickness. This phenomenon can be easily understood if we think that the horizontal distance between the margins of the tumor and the ablation zone does not reflect the thinnest gap. This discrepancy was evidenced by our results regarding comparison between quantitative and qualitative methods, which showed that 2.7% and 34.5% of the cases, respectively, were assessed by each method to establish at least 5-mm ablative margin (p < 0.0001). Considering this point, the commonly accepted 5-mm ablative margin, which is usually evaluated with only axial image sets, may correspond to the 3-mm ablative margin of our results, which was evaluated by the 3D method.
The concept of local tumor progression implies the presence of both an incompletely treated viable tumor that was previously considered to be completely treated but continued to grow and new tumor growth at the original site, such as HCC satellite tumors [1]. Residual tumor growth as a cause of local tumor progression has been confirmed by a previous study that found that the presence of viable tumor cells adherent to the radiofrequency electrode after ablation was an independent predictor of local tumor progression [12]. With the use of current imaging techniques, it is impossible to detect such residual viable tumor cells. However, when local tumor progression has been cited in clinical studies, especially with risk factor analysis, it is usually understood that the local tumor progression refers to previous growth rather than de novo growth. For example, when it is stated that the RFA of a larger tumor has a higher risk of local tumor progression than that of a smaller tumor, it is suggested that RFA of a large tumor has an increased risk for peritumoral infiltration of unablated viable tumor cells, given the same size of the ablation zone. In our study, the term “local tumor progression” generally was applied to both concepts. However, the concept of concordant local tumor progression refers to locally recurrent tumor originating from untreated tumor cells beyond the ablation zone. Our study was based on this premise. If the distances of tumor cell infiltration from the tumor surface were the same all around the tumor, concordant local tumor progression due to unablated viable tumor cells would tend to occur at the area of the thinnest ablative margin. Discordant local tumor progression due to de novo carcinogenesis or tumor cells seeded by multiple electrode replacement can occur at any site within the ablation zone, even at the thickest ablative margin.
The multivariate analysis revealed that the ablative margin was the only significant factor independently associated with local tumor progression after RFA of HCC measuring larger than 2 cm and smaller than 5 cm. The results of this study suggest that a sufficient ablative margin is more important than the presence of hepatic vessels near the tumor or a large tumor size. Although previous studies [5, 9] have shown that a hepatic vessel close to a tumor or a large tumor size were significant risk factors for local tumor progression, it is recommended that the finding be interpreted as hepatic vessel–induced indentation of the ablation zone causing an insufficient ablative margin or a large tumor size, which subsequently decreases the ablative margin, instead of the presence of the hepatic vessel or a large size itself being a significant risk factor.
Two major factors that influenced the ablative margin in this study were hepatic vessels and the geographic relationship between the tumor and ablation zone. The latter factor depends on the technical aspects of the RFA procedure, such as the choice of electrode, the number of multiple overlapping ablations, positioning of the electrode, and possibly the discrepancy between imaging techniques for guiding and assessment after RFA, which can be overcome by the operator's skill and experience [19]. However, the effects of hepatic vessels are uncontrollable. To address such problems, combination therapy has been used. Chemoembolization before RFA suppresses hepatic arterial flow and theoretically reduces the heat-sink effect. Combination with ethanol injection therapy can also be expected to lower local tumor progression rates because pretreatment with an ethanol injection has been shown to augment the RFA zone by reducing blood flow around the tumor in an animal model [20]. Several reports have confirmed the beneficial effects of these combination therapies [2124]. Second, more-aggressive ablation of the portion of the tumor close to a hepatic vessel might be useful. If the tumor portion located near a hepatic vessel is treated more aggressively—for example, by placing the electrode closer to the adjacent hepatic vessel in the tumor—as long as the ablation zone can encompass the whole tumor and margin, the treatment results would be improved by destroying small vessels or at least counteracting the heat-sink effect.
Only HCC tumors more than 2 cm in diameter were included and evaluated in this study because the incidence of local tumor progression is known to correlate with the size of the tumor [2, 3, 5, 6]; thus, the clinical interest regarding local tumor progression is much greater in cases with larger HCC tumors. In addition, it was desirable from a methodologic perspective because the criteria for diagnosing an HCC without biopsy are less strict when the tumor is larger than 2 cm, according to the American Association for the Study of Liver Diseases criteria [16].
For the 3D analysis of ablative margins, image fusion and radial multiplanar reformation techniques were used in this study. The combination of both techniques enabled us to evaluate the ablative margin quantitatively. The use of the radial multiplanar reformation technique made possible a thorough examination in virtually all directions. If image fusion or the radial multiplanar reformation technique is unavailable, utilization of either technique—for instance, side-by-side evaluation of radial multiplanar reformation images or image fusion of conventional coronal or sagittal multiplanar reformation images—can be useful.
During the quantitative analysis of the thinnest ablative margin, we ignored the restrictions of the ablation zone caused by the hepatic capsule. We postulated that there was no space where tumor cell infiltration spread beyond the ablation zone as far as the ablation reached to the hepatic parenchyma just beneath the capsule. This was confirmed by no local tumor progression at the subcapsular ablative margin, even in cases where the subcapsular HCC tumor abutted the hepatic capsule.
There are some limitations to this study. First, because data were collected retrospectively, a number of ablations were excluded. Second, there were some technical problems in the quantitative analysis. Because only a rigid registration method was available, it was difficult in some cases to synchronize the two sets of CT images completely, because the liver is not a rigid organ. In cases in which the discrepancy of the liver findings was not very significant, we were able to register images giving priority to the area around the tumor, although certain misregistrations occurred in other parts of the liver. However, as a result, we were unable to assess 23 (18.3%) of the 126 tumors for which the appropriate CT studies were available, even after using this strategy. A nonrigid registration algorithm that can overcome the so-called “tissue slide effect” is needed [2527] to address these problems. Third, in the evaluation of concordance between local tumor progression and the thinnest ablative margin, a difference within ± 1 o'clock was regarded as being concordant. However, such evaluation might induce overestimation of concordant local tumor progression, or vice versa. Finally, the results of our study were not confirmed by histologic analysis.
In conclusion, the feasibility of 3D quantitative analysis of ablative margins after RFA of HCC using CT image fusion and radial multiplanar reformation techniques has been validated. Hepatic vessel–induced indentation of the ablation zone was an important cause of an insufficient ablative margin. The failure of establishing at least a 3-mm ablative margin was the only significant risk factor associated with concordant local tumor progression. However, the 3-mm margin might be used as a criterion for more elaborate attention or closer follow-up rather than a goal of ablation because there is no doubt that a large ablative margin lowers the rate of local tumor progression.

Footnote

Address correspondence to W. J. Lee ([email protected]).

References

1.
Goldberg SN, Grassi CJ, Cardella JF, et al.; Society of Interventional Radiology Technology Assessment Committee; International Working Group on Image-Guided Tumor Ablation. Image-guided tumor ablation: standardization of terminology and reporting criteria. Radiology 2005; 235:728–739
2.
Choi D, Lim HK, Rhim H, et al. Percutaneous radiofrequency ablation for early-stage hepatocellular carcinoma as a first-line treatment: long-term results and prognostic factors in a large single-institution series. Eur Radiol 2007; 17:684–692
3.
Hori T, Nagata K, Hasuike S, et al. Risk factors for the local recurrence of hepatocellular carcinoma after a single session of percutaneous radiofrequency ablation. J Gastroenterol 2003; 38:977 –981
4.
Kim SH, Lim HK, Choi D, et al. Percutaneous radiofrequency ablation of hepatocellular carcinoma: effect of histologic grade on therapeutic results. AJR 2006; 186 [suppl 5]:S327 –S333
5.
Kim YS, Rhim H, Cho OK, Koh BH, Kim Y. Intrahepatic recurrence after percutaneous radiofrequency ablation of hepatocellular carcinoma: analysis of the pattern and risk factors. Eur J Radiol 2006; 59:432 –441
6.
Komorizono Y, Oketani M, Sako K, et al. Risk factors for local recurrence of small hepatocellular carcinoma tumors after a single session, single application of percutaneous radiofrequency ablation. Cancer 2003; 97:1253 –1262
7.
Lencioni R, Cioni D, Crocetti L, et al. Early-stage hepatocellular carcinoma in patients with cirrhosis: long-term results of percutaneous image-guided radiofrequency ablation. Radiology 2005; 234:961 –967
8.
Livraghi T, Meloni F, Di Stasi M, et al. Sustained complete response and complications rates after radiofrequency ablation of very early hepatocellular carcinoma in cirrhosis: is resection still the treatment of choice? Hepatology 2008; 47:82–89
9.
Nakazawa T, Kokubu S, Shibuya A, et al. Radiofrequency ablation of hepatocellular carcinoma: correlation between local tumor progression after ablation and ablative margin. AJR 2007; 188:480–488
10.
Izumi N, Asahina Y, Noguchi O, et al. Risk factors for distant recurrence of hepatocellular carcinoma in the liver after complete coagulation by microwave or radiofrequency ablation. Cancer 2001; 91:949 –956
11.
Okuwaki Y, Nakazawa T, Shibuya A, et al. Intrahepatic distant recurrence after radiofrequency ablation for a single small hepatocellular carcinoma: risk factors and patterns. J Gastroenterol 2008; 43:71 –78
12.
Sofocleous CT, Nascimento RG, Petrovic LM, et al. Histopathologic and immunohistochemical features of tissue adherent to multitined electrodes after RF ablation of liver malignancies can help predict local tumor progression: initial results. Radiology 2008; 249:364 –374
13.
Chau GY, Lui WY, Tsay SH, et al. Prognostic significance of surgical margin in hepatocellular carcinoma resection: an analysis of 165 Childs' A patients. J Surg Oncol 1997; 66:122–126
14.
Poon RT, Fan ST, Ng IO, Wong J. Significance of resection margin in hepatectomy for hepatocellular carcinoma: a critical reappraisal. Ann Surg 2000; 231:544–551
15.
Yoshida Y, Kanematsu T, Matsumata T, Takenaka K, Sugimachi K. Surgical margin and recurrence after resection of hepatocellular carcinoma in patients with cirrhosis: further evaluation of limited hepatic resection. Ann Surg 1989; 209:297–301
16.
Bruix J, Sherman M. Practice Guidelines Committee, American Association for the Study of Liver Diseases. Management of hepatocellular carcinoma. Hepatology 2005; 42:1208 –1236
17.
Lim HK, Choi D, Lee WJ, et al. Hepatocellular carcinoma treated with percutaneous radio-frequency ablation: evaluation with follow-up multiphase helical CT. Radiology 2001; 221:447–454
18.
Rhim H, Lim HK, Kim YS, Choi D. Percutaneous radiofrequency ablation with artificial ascites for hepatocellular carcinoma in the hepatic dome: initial experience. AJR 2008; 190:91–98
19.
Rhim H, Goldberg SN, Dodd GD, 3rd, et al. Essential techniques for successful radio-frequency thermal ablation of malignant hepatic tumors. RadioGraphics 2001; 21:S17–S35; discussion, S36–S39
20.
Goldberg SN, Kruskal JB, Oliver BS, Clouse ME, Gazelle GS. Percutaneous tumor ablation: increased coagulation by combining radio-frequency ablation and ethanol instillation in a rat breast tumor model. Radiology 2000; 217:827–831
21.
Sakr AA, Saleh AA, Moeaty AA, Moeaty AA. The combined effect of radiofrequency and ethanol ablation in the management of large hepatocellular carcinoma. Eur J Radiol 2005; 54:418–425
22.
Wong SN, Lin CJ, Lin CC, Chen WT, Cua IH, Lin SM. Combined percutaneous radiofrequency ablation and ethanol injection for hepatocellular carcinoma in high-risk locations. AJR 2008; 190:623; [web]W187–W195
23.
Yamakado K, Nakatsuka A, Takaki H, et al. Early-stage hepatocellular carcinoma: radiofrequency ablation combined with chemoembolization versus hepatectomy. Radiology 2008; 247:260 –266
24.
Zhang YJ, Liang HH, Chen MS, et al. Hepatocellular carcinoma treated with radiofrequency ablation with or without ethanol injection: a prospective randomized trial. Radiology 2007; 244:599 –607
25.
Noblet V, Heinrich C, Heitz F, Armspach JP. 3-D deformable image registration: a topology preservation scheme based on hierarchical deformation models and interval analysis optimization. IEEE Trans Image Process 2005; 14:553 –566
26.
Rao A, Chandrashekara R, Sanchez-Ortiz GI, et al. Spatial transformation of motion and deformation fields using nonrigid registration. IEEE Trans Med Imaging 2004; 23:1065 –1076
27.
Rohlfing T, Maurer CR Jr, Bluemke DA, Jacobs MA. Volume-preserving nonrigid registration of MR breast images using free-form deformation with an incompressibility constraint. IEEE Trans Med Imaging 2003; 22:730 –741

Information & Authors

Information

Published In

American Journal of Roentgenology
Pages: 758 - 765
PubMed: 20729457

History

Submitted: April 23, 2009
Accepted: February 17, 2010
First published: November 23, 2012

Keywords

  1. ablative margin
  2. hepatocellular carcinoma
  3. local tumor progression
  4. radiofrequency ablation

Authors

Affiliations

Young-sun Kim
All authors: Department of Radiology and Center for Imaging Science, Samsung Medical Center, Sungkyunkwan University School of Medicine, 50 Irwon-dong, Gangnam-gu, Seoul 135-710, Republic of Korea.
Won Jae Lee
All authors: Department of Radiology and Center for Imaging Science, Samsung Medical Center, Sungkyunkwan University School of Medicine, 50 Irwon-dong, Gangnam-gu, Seoul 135-710, Republic of Korea.
Hyunchul Rhim
All authors: Department of Radiology and Center for Imaging Science, Samsung Medical Center, Sungkyunkwan University School of Medicine, 50 Irwon-dong, Gangnam-gu, Seoul 135-710, Republic of Korea.
Hyo K. Lim
All authors: Department of Radiology and Center for Imaging Science, Samsung Medical Center, Sungkyunkwan University School of Medicine, 50 Irwon-dong, Gangnam-gu, Seoul 135-710, Republic of Korea.
Dongil Choi
All authors: Department of Radiology and Center for Imaging Science, Samsung Medical Center, Sungkyunkwan University School of Medicine, 50 Irwon-dong, Gangnam-gu, Seoul 135-710, Republic of Korea.
Ji Young Lee
All authors: Department of Radiology and Center for Imaging Science, Samsung Medical Center, Sungkyunkwan University School of Medicine, 50 Irwon-dong, Gangnam-gu, Seoul 135-710, Republic of Korea.

Metrics & Citations

Metrics

Citations

Export Citations

To download the citation to this article, select your reference manager software.

Articles citing this article

Media

Figures

Other

Tables

Share

Share

Copy the content Link

Share on social media