Recurrence-free Survival after Radiofrequency Ablation of Hepatocellular Carcinoma. A Registry Report of the Impact of Risk Factors on Outcome

Background: Despite the high complete necrosis rate of radiofrequency ablation (RFA), tumor recurrence, either local tumor recurrence or new tumor formation, remains a signiﬁcant problem. Purpose of this study is to evaluate the pattern and risk factors for intrahepatic recurrence after percutaneous RFA for hepatocellular carcinoma (HCC). Methods: We studied 40 patients with 48 HCCs ( (cid:2) 3.5 cm) who were treated with percutaneous RFA. The mean follow-up period was 24.1 + 15.7 months. We evaluated the cumulative disease-free survival of overall intrahepatic recurrence, local tumor progression (LTP) and intrahepatic distant recurrence (IDR). Thirty host, tumoral and therapeutic risk factors were reviewed for signiﬁcant tie-in correlation with recurrence: age; gender; whether RFA was the initial treatment for HCC or not; severity of liver disease; cause of liver cirrhosis; contact of tumor to major hepatic vessels and liver capsule; degree of approximation of tumor to the liver hilum; ablation time; degree of benign pre-ablational enhancement; sufﬁcient safety margin; tumor multinodu-larity; tumor histological differentiation; tumor segmental location; maximum tumor diameter; degree of tumor pre-ablational enhancement at arterial phase CT, MRI or CT-angiography; and laboratory markers pre- and post-ablation (AFP, PIVKA II, TP, AST, ALT, ALP and TB). Results: The incidence of overall recurrence, LTP and IDR was 65, 23 and 52.5%, respectively. The cumulative disease-free survival rates were 54.6, 74.8 and 78.3% at 1 year, 27.3, 71.9 and 46.3% at 2 years and 20, 71.9 and 29.4 at 3 years, respectively. Univariate and multivariate analysis showed that the signiﬁcant risk factors for LTP were: tumor size (cid:3) 2.3 cm, insufﬁcient safety margin, multinodular tumor, tumors located at segments 8 and 5, and patient’s age . 65 years ( P , 0.05). No signiﬁcant risk factor relationship for IDR could be detected. Conclusion: Our results would have clinical implications for advance warning and appropriate management of patients scheduled for RFA. Patients at risk of LTP should be closely monitored in the ﬁrst year. Furthermore, regular long-term surveillance is essential for early detection and eradication of IDR.


INTRODUCTION
Hepatocellular carcinoma (HCC) is the cause of 250 000 deaths worldwide each year. HCC is often advanced at first manifestation, and without treatment the 5-year survival rate is less than 5% (1). Only 9 -27% of the patients with HCC are eligible for surgical resection. There are many limiting factors for successful surgical resection in patients with HCC such as severe impairment of hepatic functional reserve, bilobar distribution of the tumors, extra hepatic metastasis or involvement of the portal vein (2,3). Hence, different locoregional therapies have been developed for irresectable liver tumors with an attempt to achieve local tumor control. These include transarterial chemoembolization (TACE), percutaneous ethanol injection (PEI), and various thermal ablation therapies such as cryotherapy, interstitial laser therapy, microwave coagulation and radiofrequency ablation (4 -7). Although these ablation therapies can achieve complete necrosis of small HCC, recurrence is still common. Most HCCs are associated with liver cirrhosis, and total prevention of recurrence might not be achieved even in the future (8). The intrahepatic recurrence rate is 20% during a mean follow-up period of 18 months (9). However, it is still unclear which factors influence intrahepatic recurrence (10).
Among the local methods for tumor control, radiofrequency ablation (RFA) is considered a promising alternative to surgery (11). For irresectable tumors, RFA seems to be the most effective treatment among other locoregional therapies. The main advantages of RFA include low morbidity and mortality rates, effective tumor ablation and preservation of maximal normal liver parenchyma (12). However, despite the high complete necrosis rate of RFA, early tumor recurrence within one year, either local tumor recurrence or new tumor formation, remains a significant problem. A series of studies discussed the factors for tumor recurrence, including the tumor size, subcapsular lesion, operative procedure, underlying liver disease and alpha-fetoprotein (AFP) levels, but the results were not well documented (13 -17).
There are two types of intrahepatic recurrence found in patients with HCC after RFA, local tumor progression and intrahepatic distant recurrence. Local tumor progression (LTP) occurs along the peripheral margin of the ablative lesion and intrahepatic distant recurrence (IDR) is a new HCC tumor remote from the margin of the ablative lesion. Evaluation of overall recurrence as well as LTP and IDR may provide important information for the management of HCC patients receiving RFA therapy (18).
Therefore, the purpose of this study was to determine the pattern of cumulative recurrence-free survival rate after percutaneous RFA of hepatocellular carcinoma and to determine the risk factors and the incidence of intrahepatic recurrence according to each type. To assign that task, many potential risk factors were reviewed for significant tie-in correlation to intrahepatic recurrence.

STUDY POPULATION
From February 2000 to November 2005, a total of 40 patients with 48 HCC nodules underwent percutaneous RFA for the treatment of HCC by a multidisciplinary team consisting of hepatobiliary physicians and interventional radiologists. Ultrasound-guided ablation was performed in the Hepatobiliary Oncology Department and CT-guided ablation in the Radiology Department, National Cancer Center Hospital East, Japan.
In all patients, a written informed consent was obtained. This study was approved by our institutional review board. Ablation performed in inpatient participants after they had fasted for 6 h. Laboratory examinations including complete blood count, blood coagulation test, blood typing and tumor marker for HCC (AFP) were performed before each procedure. Inclusion criteria for performing RFA in patients with HCC are as follows: the tumor or tumors should be visualized with ultrasonography (US) or CT and accessible via the percutaneous route; a single tumor no greater than 5 cm in the largest dimension; multiple tumors (,3) with each tumor measuring no greater than 3 cm; no portal venous thrombosis and extrahepatic metastasis; prothrombin time ratio over 50% ( prothrombin time with international normalized ratio, INR , 1.7) and a platelet count greater than 50 000/ml without transfusion support.
For all patients, proof of HCC malignancy was attained by the typical imaging features at US, triphasic CT, triphasic MRI, celiac angiography or CT-angiography along with a serum AFP, PIVKA II tumor markers and US-guided biopsy. The imaging criteria for HCC were a newly presenting, residual or recurrent tumor at follow-up US or CT in patients with chronic liver disease or a characteristic enhancement pattern on contrast-enhanced multi-phase CT and/or MRI (hypervascularization on hepatic arterial phase and wash-out pattern on delayed phase), filling defect at CTAP or tumor staining at DSA and CTA.
The patients were predominantly elderly (mean age 65.6 + 8.5) and male (77.5%). Five patients had a past history of HCC therapy with transcatheter arterial chemoembolization (n ¼ 3), surgical resection (n ¼ 1) or percutaneous microwave ablation (n ¼ 1), prior to RFA for the recurrent tumor. All the tumors in these five patients included in our study were newly developed lesions, and there was no evidence of viability of the tumors that had been treated previously. Mean value of the period between the latest previous treatment and RFA was 18 + 12.8 months with a range of 3 -34 months.
Severity of the underlying cirrhosis was classified in accordance with the Child -Pugh classification. The etiology of cirrhosis was chronic hepatitis C virus (HCV) infection in 33 (82.8%) patients, hepatitis B virus (HBV) infection in four (10%) patients and cirrhosis due to hepatitis of non-viral causes in three (7.5%) patients. Clinical features of patients and tumors are summarized in Table 1.

RADIOFREQUENCY ABLATION
The RF system used in this study was the RF 2000 (RadioTherapeutics, Mountain View, CA, USA), which included a 100 W generator, a 15-gauge monopolar electrode array with 10 hook-like arms and needle electrodes of 2 and 3 cm diameter (LeVeen; RadioTherapeutics). Hot withdrawal was performed to prevent oozing and tumor seeding. When the ablation was started, we often ablated the deepest parts first and the superficial parts last to avoid Jpn J Clin Oncol 2007;37(9) 659 imaging disturbances caused by hyperechogenic gas artifacts generated from boiling tissue. The average time required for ablation ranged from 3 to 51.11 min with a mean value of 20.6 + 11.8 min. The goal of radiofrequency thermal ablation is to kill the target tumor as well as a 5 mm circumferential cuff of adjacent normal hepatic parenchyma as a safety margin. RF ablations were performed via the percutaneous route under US-guidance (n ¼ 44) or CT-guidance (n ¼ 4). Image guided ablation was performed using a real-time US (GE Logiq 5 Expert) scanner with 3.5 -5.0 MHz convex probes equipped with attachments for biopsy and electrode insertion. For CT-guided tumor ablation a Toshiba Aquilion multislice-CT (MS-CT) scanner with 16 row detector channels was used. The patients received 35 mg of pethidine hydrochloride (Opystan, Tanabe, Tokyo, Japan) intravenously before RFA for analgesia. Antibiotics were administered before and 2 -3 days after each PEI and RFA procedure. Local infiltration anesthesia was induced by using 5-10 ml of 1% lidocaine.

EVALUATION OF THERAPEUTIC EFFICIENCY
To evaluate the tumor response to ablation therapy, contrast-enhanced CT was performed 1 month after the treatment. Additional CT scans were obtained in all patients within 1 day after completion of therapy to determine whether there was any remaining malignant tissue that would require a second ablation session and also for early detection of complication. The ablation was considered a success on the basis of all of the following findings at follow-up CT: (a) no contrast enhancement detected within or around the tumor; (b) the margins of the ablation zone clear and smooth; and (c) the ablation zone extended beyond the previously estimated tumor borders.
If any residual HCC was noted, we repeated the RFA as soon as possible (n ¼ 5) and then performed a second immediate follow-up CT scan. We confirmed that there was no evidence of residual tumor that went untreated in all patients. As a follow-up rule thereafter, contrast-enhanced CT, US scan and measurement of serum AFP were performed regularly every 3 -4 months.

RISK FACTORS FOR RECURRENCE
Intrahepatic recurrence was divided into LTP and IDR based on standardization of terminology and reporting criteria by the international working group of image-guided tumor ablation (19). Local tumor recurrence of an HCC nodule was defined as the development of an enhanced area on the CT scan in the same sub-segment as the primary nodule and was found along the peripheral margin of the ablative zone. Intrahepatic distant recurrence of HCC was defined as a lesion with typical enhancement characteristics for HCC but distant from the original ablative zone. IDR was evaluated for patients for whom complete coagulation was achieved without recurrence in the same sub-segment as the primary nodule. In case of recurrence, other supplemental examinations like MRI, hepatic DSA and angio-CT were performed to confirm not only local but also distant recurrence of HCC. When tumor recurrence was confirmed, patients were hospitalized and an additional treatment cycle was administered if the patient's physical condition was strong enough for him or her to safely undergo another ablation session(s). Patients who developed diffuse HCC were shifted to transarterial embolization (TAE), hepatic artery infusion (TAI), radiotherapy or proton beam therapy.

STATISTICAL ANALYSIS
After review of the follow-up CT, we assessed the incidence and the cumulative recurrence-free survival rate of overall recurrence, LTP and IDR, respectively. Data analyses were performed using MedCalc statistical software for Biomedical research (version 9 for Windows).
For analysis of the significant risk factors for LTP, 30 host, tumoral and therapeutic variables were reviewed: (1) age, (2) gender, (3) whether RFA was the initial treatment

660
Recurrence-free survival after RFA of HCC for HCC or not (previous history of interventional or surgical), (4) severity of liver disease according to standard Child -Pugh classification, (5) cause of liver cirrhosis (hepatitis C or B or non viral hepatitis), (6 and 7) contact of tumor to major hepatic vessels and liver capsule, (8) degree of approximation of tumor to the liver hilum, (9) ablation time, (10) the degree of benign periablational enhancement at immediate follow-up CT (grade 0, no enhancement; grade 1, disrupted non-continuous enhancement; grade 2, continuous rim enhancement), (11) ablation safety margin is sufficient or not (sufficient 5 mm, pre-and post ablation CT were as reference), (12) number of primary HCC nodules at the time of ablation (single or multiple), (13) tumor histological differentiation (well, moderate or poorly differentiated), (14) tumor segmental location (in segments 5 and 8 or others), (15) maximum tumor diameter (we divided the tumors into small and large groups with the cutting value of 2.3 cm), (16) degree of tumor pre-ablational enhancement at arterial phase CT, MRI or CT-angiography (strongly or poorly enhanced), (17 -30) laboratory markers pre-and postablation within normal limits or not (alpha-fetoprotein, desgamma carboxy prothrombin, total protein, aspartate transaminase enzyme, alanine transaminase enzyme, alkaline phosphatase enzyme and total bilirubin). For the IDR, 21 host, tumoral and therapeutic variables were reviewed: (1) age, (2) gender, (3) whether RFA was the initial treatment for HCC or not (previous history of interventional or surgical), (4) severity of liver disease according to standard Child-Pugh classification, (5) cause of liver cirrhosis (hepatitis C or B or non viral hepatitis), (6) number of primary HCC nodules at the time of ablation (single or multiple), (7) tumor histological differentiation (well, moderate, or poorly differentiated), (8-21) laboratory markers pre-and post-ablation within normal limits or not (alpha-fetoprotein, des-gamma carboxy prothrombin, total protein, aspartate transaminase enzyme, alanine transaminase enzyme, alkaline phosphatase enzyme and total bilirubin). These risk factors were analyzed retrospectively.
The unpaired Student's t-test was used to compare averages between groups and the x2-test and Fisher's exact probability test were used to compare independence. Cumulative disease-free survival was estimated using the Kaplan -Meier method and the significance of the hazard ratio for LTP and IDR was evaluated with univariate analysis using the log-rank test. If multiple hazard ratios were proven to be significant by this test, we performed multivariate analysis using a stepwise Cox proportional hazard regression model to search for independently significant risk factors. The results were reported as ratios with 95% CI. A P-value ,0.05 was considered statistically significant.
Statistical analysis determined high hazard ratio for elevated AFP post-ablation (2.476) and low total protein (TP) pre-ablation (2.411); however, the P-value was not significant for both (P ¼ 0.2368 and 0.3946 respectively).
We could not determine a significant correlation between the incidence of LTP and gender, previous HCC therapy, severity of liver disease, cause of liver cirrhosis, contact of tumor to major hepatic vessels and liver capsule, degree of approximation of tumor to liver hilum, ablation time, the

662
Recurrence-free survival after RFA of HCC degree of benign pre-ablational enhancement, tumor histological differentiation, degree of tumor enhancement preablation or laboratory markers. Also the hazard ratios for all were low, except for elevated AFP post-ablation and low TP pre-ablation.

INTRAHEPATIC DISTANT RECURRENCE
IDR occurred in 21 of 40 patients (52.5%). Among the 21 risk factors investigated for IDR, no significant correlation could be described with the length of the IDR-free survival. The highest hazard ratio (2.274) was recorded for elevated AFP pre-ablation in spite of the correlation not being significant (P ¼ 0.0777; Table 5).

DISCUSSION
We categorized intrahepatic recurrence of HCC after percutaneous RFA into local tumor progression and intrahepatic distant recurrence; each type of recurrence has a specific mechanism of pathogenesis and is thought to occur independently. Generally, LTP occurs early and is considered to be related to residual tumor cells that have spread microscopically beyond the ablative margin, although there is a possibility of de novo occurrence at that site. Pathogenesis of IDR is thought to be the result of an intrahepatic metastasis of a primary HCC or due to a multicentric origin of the HCC (20,21). Therefore, LTP may be associated more with a treatment methodology or result, local environment of the tumor such as a vessel contact and characteristics of the tumor itself rather than the systemic condition of the patient. By contrast, IDR may be related more to systemic factors rather than local factors (18). Therefore, we analyzed a variety of potential local and systemic risk factors for LTP and IDR associated with HCC, independently.
There have been several studies reporting on the incidence and risk factors of LTP or IDR after RFA for HCC. Komorizono et al. (13), who studied LTP after a single application of RF energy for relatively small HCC, reported a tumor-free survival rate at 12 and 15 months of 76 and 74%, respectively. Significant risk factors for LTP were reported to be a large tumor size over 2 cm in the greatest dimension and a subcapsular location. Hori et al. (22), who also studied a similar group of subjects, reported that the cumulative local    (18), who reported that overall recurrence rate after RFA is two-thirds of treated patients; IDR occurs more frequently than LTP and LTP almost always occurs during the first 24 months postablation; IDR could occur at any time during the first 24 months postablation or later on (Table 6).
In two patients included in our study, residual tumors were found on immediate follow-up CT scans. In those cases, we repeated the procedures as soon as possible to achieve complete ablation. Then, we confirmed complete treatment on another immediate follow-up CT scan after the additional RFA. We included these cases in our study population and analyzed them as the same as those cases with a single complete ablation; this is because there was no reason to differentiate them from the other cases with a singlesession ablation.
Hori et al. (22) reported that recurrence is low in HCC 2.3 cm and increased if the tumor is 2.5 cm in its maximum diameter. Our results in harmony with Hori et al. Reported in other series as well as our study, that the most important variable which influences the LTP, is tumor size (10,22). In the present study, large tumor size (larger than or equal to 2.3 cm) proved the highest risk ratio for LTP (8.4036, P ¼ 0.0096). Matching our results is the large study conducted by Nakashima et al. (33). Nakashima et al. investigated the relationship between macroscopic types of HCC and intrahepatic metastasis in 209 surgically resected small HCC nodules less than 3 cm in diameter. Vaguely nodular and single nodular types (mean diameter 1.36 and 2.28 cm, respectively) were found to have lower prevalence of intrahepatic metastasis (0 and 4.1%, respectively) than single nodular with extranodular growth and confluent multinodular types (mean diameter 2.31 and 2.39 cm, intrahepatic Recurrence-free survival after RFA of HCC    (33). It was reported that the coagulated necrotic area produced by RFA conformed to the size of the tumor and was smaller than expected in the surrounding cirrhotic tissue, and larger than expected within the tumor (called the 'oven effect') (24). In the present study, 16 of 48 tumors were 2.3 cm in diameter with LTP incidence 8/16 (50%). Several small satellite nodules might have existed around tumors larger than 25 mm in diameter and these satellite nodules could not be completely treated by single-session RFA in tumors larger than 25 mm since they are difficult to detect by transcutaneous US, dynamic CT or dynamic MRI (22,25).
Although there have been studies on a safe tumor-free margin of a surgical resection for hepatic tumors, this remains an unresolved problem (21,26,27). A gross tumorfree margin of 1 cm in all directions around the tumor in the resected specimen is generally accepted by most surgeons and pathologists. Local recurrence was more frequently observed after limited resections than after anatomic resections that included the tumor and its portal territory (50 vs 10%). Patients undergoing anatomic resection for HCC achieved better disease-free survival than those undergoing limited resection (26). RFA in our study was performed percutaneously; evaluation could be done with CT images preand post-procedure. We set and evaluated a safe ablative margin as 5 mm in all directions around the tumor and two CT scans before and after the RFA were analyzed. We showed that the establishment of a 5 mm ablative margin was effective in suppressing LTP after RFA of HCC and, without this safety margin ablation, LTP incidence was 57.1% (risk ratio ¼ 6.2643, P ¼ 0.0396).
Tumor multiplicity is a strong indicator of tumor recurrence after curative resection of HCC (8), microwave (23) or radiofrequency ablation (10,23). We obtained the same end result; statistical analysis of the number of primary HCC nodules at the time of ablation revealed that the LTP incidence was 7/14 (50%) for primary multinodular HCC. The primary multinodular HCC was significantly associated with a higher LTP (risk ratio ¼ 5.1790, P ¼ 0.0482).
Tumor location is one of the most important factors influencing LTP. Eleven LTP tumors showed segmental distribution as follows: S8 (n ¼ 4), S5 (n ¼ 2), S8 þ 5 (n ¼ 3) and S6 (n ¼ 2). One of the reasons why segment 8 þ 5 showed a higher LTP rate may be that there were more chances for dual arterial feeders from segmental arteries. The other possible reason may be the greater probability for collateral vessels developing from the adjacent segmental arteries after ablation. This agrees with the conclusion of Yun et al. (28), concerning significant risk for local recurrence after chemoembolization for HCC in tumor located in segmental border zone. If tumor located in S8 attained a sub-diaphragmatic location, then LTP was significantly higher compared with that in tumors situated more deeply within the liver parenchyma. When the tumors were located close to the liver surface beneath the diaphragm, it was not easy to insert the RFA electrode and open the multiple-array at the center of the tumor. In the present study, three of four S8 locally recurrent tumors were located beneath the diaphragm. Recently, it was reported that the laparoscopic approach and artificial ascites method for percutaneous treatment were effective techniques for the treatment of patients with HCC located just beneath the diaphragm (24,29 -31). These techniques are recommended for the treatment of tumors located close to the surface (22). In the same way, tumors in segment 5 have a greater chance of being near to the gall bladder and consequently in this location it is complicated to achieve complete ablation by RFA. In this study, one of two tumors at S5 was located near to the gall bladder. Percutaneous ethanol injection was recommended in such patients (risk ratio ¼ 4.6420, P ¼ 0.0007).
The patient's age was confirmed as an independent prognostic variable, perhaps representing a surrogate for declining host defense mechanism associated with advancing age. The incidence of LTP was 9/27 (33.3%) in patients .65 years old (risk ratio 4.6420, P ¼ 0.0011). Theoretically, a tumor that is contiguous to a large vessel has more chance of allowing some tumor cells to survive local thermal therapy because there is a significant tissue cooling effect caused by blood circulation of normal body temperature (32). However, previous studies (13,18,22) showed that it was not a significant risk factor for LTP after RFA for HCC and our results agree with them; we could not ascertain that contiguity to a large vessel is a significant risk factor for LTP.
Previous studies proved that an increased level of serum AFP was associated with IDR after RFA for HCC (14,16,18,23). In our study the correlation between serum AFP and IDR was not significant (P ¼ 0.0777); however the pre-ablation elevated AFP level recorded the highest hazard ratio (2.274) among all other risk factors investigated for IDR.
In our study, 91% of LTP occurred after 12 months and after 18 months; tumors that did not recur during that interval did not show any LTP during the residual follow-up period of up to 50 months. This time interval has great clinical significance because it can affect the follow-up CT schedule as well as patient prognosis. It is an important explanation for why the close follow-up protocol is indispensable during the first year. Kim et al. (18) support our standpoint; in his study, the incidence of LTP after 24 months is 0%.

CONCLUSION
We can conclude that, after percutaneous RFA for HCC, overall recurrence rate after RFA reaches two-thirds of the treated patients. IDR occurs more frequently than LTP and LTP almost always occurs during the first 18 months postablation; IDR could occur at any time earlier during the first 24 months post-ablation or later on. Although less frequent, LTP tended to occur when we ablated a large HCC tumor 2.3 cm in dimension, or when we could not establish a sufficient safety margin. Tumor multinodularity, tumors located at segments 8 and 5, and patients over 65 years are further significant risk factors. Our findings of different risk factors and prognostic factors for intrahepatic recurrence after RFA of HCC may have clinical implications in determining rational strategies in post-ablation surveillance, prevention and management of recurrence. Patients at risk of LTP should be closely monitored in the first year. Furthermore, regular long-term surveillance is essential for early detection and eradication of IDR.