Radiofrequency Ablation for HepatocellularCarcinoma: A Prospective Comparison of FourRadiofrequency DevicesShi-Ming Lin, MD, Chen-Chun Lin, MD, Wei-Ting Chen, MD, Yi-Chen Chen, MD, and Chao-Wei Hsu, MD

PURPOSE: To compare the effectiveness of ablation techniques for hepatocellular carcinoma (HCC) with the use offour radiofrequency (RF) devices.

MATERIALS AND METHODS: One hundred patients with 133 HCC lesions no larger than 4 cm were treated withone of four RF devices: RF 2000 (maximum power, 100 W) and RF 3000 generators (maximum power, 200 W) withLeVeen expandable electrodes with a maximum dimension of 3.5 cm or 4 cm, internally cooled single electrode witha thermal dimension of 3 cm, and a RITA RF generator with expandable electrodes with a maximum dimension of5 cm.

RESULTS: Numbers of RF sessions needed per HCC to achieve complete necrosis were 1.4 � 0.5 with the RF 2000device and greater than 1.1 � 0.3 with the other three devices (P < .05). The RF 2000 device required a more interactivealgorithm than the RF 3000 device. Session times per patient were 31.7 minutes � 13.2 in the RF 2000 group and longerthan 16.6 minutes � 7.5 in the RF 3000 group, 28.3 minutes � 12 in the RITA device group, and 27.1 minutes � 12 withthe internally cooled electrode device (P < .005 for RF 2000 vs other devices and for RF 3000 vs RITA or internallycooled electrode device). Complete necrosis and local tumor progression rates at 2 years in the RF 2000, RF 3000, RITA,and internally cooled electrode device groups were 91.1%, 97.1%, 96.7%, and 96.8% and 12%, 8%, 8.2%, and 8.3%,respectively (P � .37).

CONCLUSIONS: Ablation with the RF 3000 device required a shorter time than the other three devices and requireda less interactive algorithm than the RF 2000 device. However, complete necrosis and local tumor progression rateswere similar among devices.

J Vasc Interv Radiol 2007; 18:1118–1125

Abbreviations: HCC � hepatocellular carcinoma, RF � radiofrequency

RADIOFREQUENCY (RF) ablation hasbeen developed as an effective localablation therapy for hepatocellularcarcinoma (HCC) and metastatic livercancers (1–12). Four RF devices havebeen used worldwide: RF 2000 (100 W;Radiotherapeutics, Mountain View,Calif) and RF 3000 (200 W; Radiother-

From the Liver Research Unit, Chang Gung Memo-rial Hospital, Chang Gung University, 199 TungHwa North Road, Taipei, Taiwan. Received Decem-ber 5, 2006; final revision received May 26, 2007;accepted June 2, 2007. Address correspondence toS.M.L.; E-mail: lsmpaicyto@cgmh.org.tw

None of the authors have identified a conflict ofinterest.

© SIR, 2007

DOI: 10.1016/j.jvir.2007.06.010

1118

apeutics) generators with expandableLeVeen electrodes (4,6–8), a generatorwith an expandable electrode (RITAMedical Systems, Mountain View,Calif) (2,9), and a generator with aninternally cooled single electrode (5,7–12) (Radionics, Burlington, Mass). Themaximum RF power, monitoring ofimpedance and/or temperature dur-ing RF ablation, and endpoint of theRF procedure have been differentamong the four devices, so their effec-tiveness in terms of complete tumornecrosis, local tumor progression, andablation duration might be different.

The rate of complete tumor necrosisranges from 80% to 100% with indi-vidual RF devices (1–13). To our

knowledge, only two human studies

comparing RF 2000 and internallycooled single-electrode devices havebeen conducted, and both showedcomparable rates of complete tumornecrosis (8) and local tumor progres-sion (8,14). However, the first study (8)also showed a higher complicationrate in the RF 2000 group (7% vsno complications with the internallycooled RF device). No single prospec-tive study has compared the techniqueeffectiveness, ablation duration, andcomplications among the other threedevices. Hence, this prospective studywas conducted by sequentially allocat-ing patients to undergo RF ablationwith a LeVeen electrode with RF 2000or RF 3000 generators, an internally

cooled electrode, or a RITA electrode.

Lin et al • 1119Volume 18 Number 9

MATERIALS AND METHODS

Study Design and Characteristics ofPatients

RF ablation for HCC with four dif-ferent devices was approved in ourinstitution. In addition, our institutiondoes not require approval for retro-spective reports for alternate treat-ment study with similar modalities.Written informed consent was ob-tained from each patient or a familymember of the patient.

Considering the thermal diametersof 3–5 cm that can be created with thefour RF devices, we selected patientswith one to three HCC lesions 4 cm orsmaller for this study. We also did notuse three-cluster internally cooledelectrodes because they can create7-cm-diameter thermal ablation zonesand can be introduced into the tumoronly through the subcostal space or anapproach via intercostal space after re-section of one rib. At the start of thestudy, there were four commerciallyavailable RF devices in our institute:the RF 2000 and RF 3000 generators(Radiotherapeutics) with 14-gauge ex-pandable LeVeen electrodes with amaximum deployed diameters of 3.5cm and 4 cm; the RITA generatorwith a 15-gauge expandable electrode(RITA Medical Systems) with a maxi-mum deployed diameter of 5 cm andmultiple point temperature measure-ments; and the Radionics generatorwith an internally cooled single elec-trode with a thermal diameter of 3 cm(Radionics).

Between August 2002 and April2003, excluding four patients who pre-ferred the Cool-tip electrode (Radion-ics) because of the smaller-gauge elec-trode, 100 consecutive patients with133 HCCs 4 cm or smaller were as-signed into four study groups accord-ing to sequential allocation: RF 2000,RF 3000, internally cooled single elec-trode, and RITA RF device. All fourdevices were available throughout theentire study period. The primary tech-nique effectiveness endpoint of thestudy was complete tumor necrosisand the secondary technique effective-ness endpoint included local tumorprogression, ablation duration, sur-vival, and major complications.

The criteria of enrollment were asfollows: (i) adult patients with one to

three HCC lesions measuring 4 cm or

smaller in diameter each; (ii) HCC lo-cated at least 0.5 cm away from theliver surface, hepatic hilum, sono-graphically visible bile duct, gallblad-der, and gastrointestinal tract; (iii)absence of vascular invasion or extra-hepatic metastasis; (iv) liver cirrhosisclassified as Child-Pugh class A/B;(v) prothrombin time 3 seconds lessthan control value; (vi) platelet countgreater than 50,000/mm3; and (vii) noprevious HCC treatment.

HCC was diagnosed by pathologicor cytologic studies in all patients;however, both examinations were notundertaken in patients with additionaltumors when their dynamic computedtomographic (CT) findings were simi-lar to those diagnosed by pathologicor cytologic analysis. All target tumorswere visible on ultrasonography (US)and tumor diameters were determinedaccording to the largest dimensionmeasured on US. The number of tu-mors was determined by CT duringarterioportography. The four groupsexhibited no significant differences interms of age, sex, etiology of underly-ing liver disease, Child-Pugh class,number of tumors, diameter of largesttumor, or histocytologic tumor grade(Table 1).

Conventional liver biochemical tests,prothrombin time, and complete bloodcell counts were measured before treat-ment. Routine power color Doppler USimaging (ProSound 5500; Aloka, Tokyo,Japan; or Toshiba 6000; Toshiba, Tokyo,Japan) was used to detect any vessel orarteriovenous shunt within or sur-rounding the target tumors. A three-phase helical CT imaging study wasused to detect any enhancement of thetumor.

Medication and Care before RFAblation

RF ablation was performed at theinpatient base as described previously(1–12). After 4 hours of fasting, meper-idine hydrochloride (Demerol; Sanofi-Synthelabo, Zuellig, France) and mi-dazolam hydrochloride (Dormicum; F.Hoffmann-La Roche, Basel, Switzer-land) were administered intravenously3–5 minutes before RF ablation. Themaximum doses of meperidine andmidazolam during the procedure were1 mg/kg and 0.1 mg/kg, respectively.However, additional doses of both

drugs were administered if severe

pain interfered with the ablation pro-cedure. Local anesthesia was estab-lished with 10 mL of 2% lidocaine.Grounding pads were applied accord-ing to manufacturer recommendation.

RF 2000 with LeVeen Electrode

RF 2000 generators included a 480-kHz generator capable of producing amaximum power of 100 W. Under USguidance, a 12- or 15-cm-long, 14- or15-gauge electrode (LeVeen electrode;Radiotherapeutics) was introducedpercutaneously into the tumor. Thefully deployed tines of the electrodewere then deployed into the tissuethat encompassed the tumor. The fullydeployed tines of this electrodeformed an array with a diameter of2–3.5 cm.

RF ablation was performed in thesecases according to the manufacturer’sstandard algorithm or an interactivealgorithm. The standard algorithmwas begun with the power set at 30–50W, increased by 10 W per minute untila maximum power of 75–100 W, andmaintained until 15 minutes of appli-cation time had elapsed or the imped-ance had decreased rapidly and the RF2000 generator’s power shut off.

RF energy was then applied in asecond phase until a second powershutoff was achieved or 10 minutes oftreatment had elapsed. Consequently,RF ablation was applied for no morethan 25 minutes in each deployment ofthe electrode.

When applying the interactive algo-rithm (15), the initial power settingswere similar to those of the standardalgorithm, but whenever the imped-ance had not changed after 8 minutesduring the first phase of RF energyapplication or 4 minutes during thesecond application phase, the tines ofthe electrode were retracted 0.5–1 cm.Power was increased manually on theRF generator to the highest level andthe tines were maintained in this re-tracted mode until the impedance be-gan to increase. When impedancebegan to increase, the tines were rede-ployed fully into the tumor while thepower remained at the highest leveland the procedure was continued untilthe power shut off.

For all four RF devices, the elec-trode was repositioned as required to

encompass large target tumors and

1120 • Comparison of Four RF Devices for HCC Ablation September 2007 JVIR

mimic a surgical margin of 1 cm.When the planned thermal lesion hadbeen made, the power was set at 20 Wfor needle tract ablation and the min-imally deployed electrode was gradu-ally withdrawn until the electrodereached the capsule of the liver.

RF 3000 with LeVeen Electrode

In contrast to the RF 2000 generator,the RF 3000 generator is capable ofproducing a maximum power of 200W. The tines of this electrode form anarray with a diameter as large as 4 cm.

The power was set at 60–80 W andincreased by 10 W every 30 seconds to130 W with no increase in impedancebeyond 200 � after 5 minutes at 130W, increased in 10-W increments ev-ery 30 seconds to 200 W, and main-

Table 1Comparisons of Clinical Characteristics

Characteristic

No. of patientsNo. of tumorsMean age � SD (y)Sex (M/F)Child-Pugh class

AB

Underlying liver diseaseHBVHCVHBV and HCVAlcohol

No. of tumors123

No. of patients with tumor size of1–2 cm2.1–3 cm3.1–4 cm

Mean size of main tumor � SD (cm)Edmondson tumor grade

I/IIIII/IV

Baseline laboratory parametersMean ALT � SD (IU/L)Mean platelet count � SD (�1000/mm�-fetoprotein (ng/mL)

�100100–200201–400�400

Mean follow-up � SD (months)*

Note.—There was no significant differenc* Median value in parentheses.

tained until 15 minutes of application

time had elapsed or the impedancehad increased rapidly and the powershut off. The application of the stan-dard or interactive algorithm, theplanned thermal lesion, and the elec-trode tract ablation during withdrawalof the electrode were similar to thoseof the RF 2000 generator with theLeVeen electrode.

RF Ablation with Internally CooledSingle-electrode Device

A 480-kHz RF generator deliveringa maximum power of 200 W was usedfor thermal ablation. Circuitry in thegenerator allowed for continuousmonitoring of impedance. The elec-trode has a 3-cm-long active distalpart.

During RF application, the maximal

the Four RF Treatment Groups

RF 2000 RF 3000

25 2534 35

55 � 11 54 � 1018/7 19/6

21 204 5

19 174 81 01 0

24 167 81 1

6 612 137 6

2.5 � 0.5 2.6 � 0.7

19 2015 15

92 � 64 102 � 658.7 � 1.9 8.7 � 1.6

8 115 43 29 8

20.6 � 4.1 (21) 20.7 � 3.9 (22) 20

n any characteristic among the four group

RF power can automatically prevent

the impedance from increasing 20 �greater than the baseline value. A peri-staltic pump (312FS/D variable speedpump; Watson-Maslow, Paris, France)ensured cooling of the electrode with0°C saline solution at a flow rate suf-ficient to maintain the temperature ofthe electrode at less than 25°C. At thestart of RF ablation, the power was setat 40 W for tumors 2 cm or smaller andat 60 W for tumors larger than 2 cm.The power was switched to maximumlevel 1 minute after starting at the firstpower setting (16).

During ablation, RF current wasemitted for 12 minutes per electrodeinsertion (17). At the end of RF abla-tion, the current and cooling circuitwere switched off and the electroderemained in place for 30 seconds toachieve the desired cool-down tem-

ITA Internally Cooled Electrode

25 2531 33� 9 52 � 99/6 17/8

20 195 6

19 206 40 10 0

19 186 60 1

4 716 135 5

� 0.7 2.7 � 0.6

18 2013 13

� 58 74 � 56� 1.0 8.6 � 1.6

9 104 44 28 93.2 (22) 21.0 � 3.4 (22)

LT � alanine aminotransferase.

in

R

591

2.6

893) 8.8

.9 �

e i s. A

perature of 60°C–65°C.

Lin et al • 1121Volume 18 Number 9

RF Ablation with RITA Device

A 460-kHz RF generator delivered amaximum power of 150 W and wasconnected to an electrode housingnine curved tines (Starburst XL; RitaMedical Systems). The RITA generator(model 1500) was connected to a com-puter (ThinkPad 360; IBM, Milan, It-aly) that continuously recorded thepower used, temperatures, and im-pedance curves obtained during abla-tion. The expandable electrode hadnine nickel-titanium lateral tines andfive thermocouples. The standard al-gorithm was performed, as the tineswere first expanded to 2 cm witha preselected target temperature of80°C, then advanced to 3 cm with atarget temperature of 105°C, and fi-nally extended to 5 cm with a targettemperature of 110°C. The targetedtemperature had to be maintained for7 minutes before monitoring of post-ablation temperatures with the fivethermocouples (3).

Considering the similarity of theRITA and LeVeen electrodes with ex-pandable tines, we also used the inter-active algorithm that was performedas the standard algorithm after the im-pedance was not increasing; the elec-trode was withdrawn halfway untilthe impedance increased and the elec-trodes were then fully deployed asthe temperature of 105°C–110°C wasachieved and maintained for an addi-tional 7 minutes. The 1-minute cool-down temperature of 60°C–70°C wasconsidered as the criterion of suc-cessful ablation. When the plannedthermal lesion had been made, theelectrode was removed slowly withneedle tract ablation at 20 W.

Care after RF Ablation

After the RF ablation had been per-formed, the electrode was removedand the punctured site was coveredwith a sterile dressing and compressedwith a sandbag. The patient was thensent to recover for at least 4 hours ofbed rest before being released. The pa-tients were observed for potentialcomplications for several additionaldays after RF ablation and dischargedfrom the hospital when they felt nosevere pain or when their body tem-

perature did not exceed 38°C.

Assessment of Effectiveness of RFAblation

The effectiveness of ablation wasassessed on three-phase helical CTperformed 2 weeks after one or twosessions of treatment. Helical CT wasperformed with a Prospeed plus sys-tem (General Electric Medical Sys-tems, Yokogawa, Japan) at 250 mAand 120 kVp, with 10-mm collimationand a 1:1 pitch. Nonionic contrast ma-terial was injected at a rate of 2.5–4mL/sec (100 mL total) with a powerinjector. The arterial phase began 20seconds after the injection with helicalbreath-hold acquisition, the portalphase began 75 seconds after the injec-tion with the same acquisition, and thedelayed phase began 5 minutes afterthe injection.

If the foci of nodular enhancementwere noted in the treated tumor onCT, those portions of the HCC wereconsidered viable. A new RF sessionwith an identical device was given aspart of another course of treatment,and CT was repeated 2 weeks after thenew RF session. The treatment wasconsidered to have failed if the foci ofnodular enhancement in the treatedtumor were observed on CT after threesessions of ablation. Complete tumornecrosis was defined as persistent hy-poattenuation of the tumor on CT 2.5months after the last ablation therapy(15). When the index tumor showedhypovascular appearance on dynamicCT, the technique’s effectiveness wasdetermined by the maximum dimen-sion of ablation on CT. When the max-imum dimensions of ablation weremeasured to be 0.5–1 cm larger thanthe original dimension before ablation,it was considered complete ablation.In addition, we also investigated therelationship between complete abla-tion and tumor size (�3 cm vs �3 cm),as well as the achievement of technicalsuccess endpoints of maximum im-pedance with power rolloff withLeVeen electrodes or a target temper-ature greater than 60°C–65°C at 30seconds after ablation with the RITAelectrode or internally cooled elec-trode. Serum �-fetoprotein was alsomeasured 2 weeks after ablation.When the baseline level was abnor-mal, it was checked again every 2–4weeks.

The assessment of long-term out-

come in terms of local tumor progres-

sion, new HCC recurrence, overallsurvival, and cancer-free survival be-gan at the time of complete tumornecrosis. Long-term follow-up exami-nations included monitoring serum�-fetoprotein levels and assessmentwith US and helical CT every 2months for the first 6 months after RFablation. Thereafter, such follow-upexaminations were conducted every 3months. Local tumor progression ofHCC was defined by the appearanceof a newly enhancing tumor on CTduring follow-up that was contiguouswith the zone that had been consid-ered completely ablated (1,6,15,18).

Statistical Analysis

A two-tailed Student t test was per-formed to compare patients who re-ceived the four methods of RF abla-tion. The �2 or Fisher exact tests wereused to compare the proportionsamong the four groups. Local tumorprogression was estimated with theKaplan-Meier method and differencesbetween groups were compared withthe log-rank test. A P value less than.05 was considered statistically signif-icant.

RESULTS

Complete Tumor Necrosis

Failure to achieve complete tumornecrosis after three courses of treat-ment was encountered in two tumorsin the RF 2000 group and one each inthe RF 3000, RITA, and internallycooled electrode groups, respectively.Complete necrosis rates were 91.1% inthe RF 2000 group, 97.1% in the RF3000 group, 96.7% in the RITA group,and 96.8% in the internally cooledelectrode group. For the 12 of 133 tu-mors (9%) that had a hypovascular ap-pearance on dynamic CT before abla-tion, all 12 tumors showed completenecrosis on CT because their maxi-mum dimensions of ablation were0.5–1 cm larger than the original di-mensions before ablation. No signifi-cant differences in the rate of completenecrosis existed among the fourgroups (two-tailed Student t test).Complete necrosis was achieved in 26of 27 tumors 3 cm or smaller (96.2%)and six of seven tumors larger than 3

cm (85.7%) in the RF 2000 group; all 29

com

1122 • Comparison of Four RF Devices for HCC Ablation September 2007 JVIR

tumors (100%) and five of six tumors(83.3%), respectively, in the RF 3000group; 27 of 28 tumors (96.4%) andfour of five tumors (80%), respectively,in the internally cooled electrodegroup; and 24 of 25 tumors (96%) andfive of six tumors (83.3%), respec-tively, in the RITA group. No signifi-cant difference in the rate of completenecrosis existed between tumors 3 cmor smaller and those larger than 3 cmin any individual group. Local tumorprogression occurred in one of 26 tu-mors (3.8%) 3 cm or smaller and twoof six tumors (33.3%) larger than 3 cmin the RF 2000 group (P � .079, Fisherexact test); one of 29 tumors (3.4%) andone of five tumors (20%), respectively,in the RF 3000 group (P � .26); none of28 tumors (0%) and two of four tumors(50%), respectively, in the internallycooled electrode group (P � .019); andnone of 25 tumors (0%) and two of fivetumors (40%), respectively, in theRITA group (P � .023). There wasmore local tumor progression in tu-mors larger than 3 cm than in tumors 3cm or smaller in the RF 2000 group,internally cooled electrode group, andRITA group. Regarding the relation-ship between complete necrosis andthe technique endpoints, the maxi-mum impedance was greater in 32 tu-mors in which complete necrosis wasachieved (mean � SD, 571 � � 77)than in two tumors in which completenecrosis was not achieved (109 � � 12;

Table 2Factors Related to Achievement of Comp

Factor

No. of patientsNo. of tumorsTreatment courses

12

Mean treatment sessions per tumor �SD (range)*

Time per patient†RF algorithm

InteractiveStandard

Mean hospital stay (days) after start ofablation � SD (range)

Mean follow-up (months) � SD (range)

Note.—This table excludes patients witho* P � .05: RF 2000 versus the other three† P � .005: RF 3000 versus the other thre‡ P � .05: RF 2000 versus RF 3000; other

P � .001) in the RF 2000 group. Max-

imum impedance was also greater inthe 34 tumors in which complete ne-crosis was achieved (640 � � 122) thanin the one tumor in which completenecrosis was not achieved (142 � � 0;P � .001). The target temperature washigher in the 31 tumors in which com-plete necrosis was achieved (75.9°C �7°C) than in the two tumors in whichcomplete necrosis was not achieved(51°C � 14°C; P � .002) in the inter-nally cooled electrode group and wasalso higher in the 29 tumors in whichcomplete necrosis was achieved(68.6°C � 2.9°C) than in the two tu-mors in which complete necrosis wasnot achieved (53°C � 1°C; P � .001) inthe RITA electrode group. Completetumor necrosis was achieved in mosttumors after one course of treatment(Table 2).

Seven tumors in the RF 2000 groupand two each in the RF 3000, RITA,and internally cooled electrode groupsrequired two sessions to achieve com-plete necrosis. Two tumors in the RF2000 group and one in the internallycooled electrode group required threesessions of RF ablation but still did notshow complete tumor necrosis be-cause of the tumor’s location near avessel exceeding 3 mm in diameter(Figure). The number of sessions of RFablation required to achieve completetumor necrosis was higher in the RF2000 group (1.4 � 0.6) than in the otherthree groups (P � .05, two-tailed Stu-

te Tumor Necrosis

RF 2000 RF 3000

23 2432 34

16 227 2

.4 � 0.5 (1–2) 1.1 � 0.3 (1–2) 1.1

.7 � 13.2 16.6 � 7.5 28.3

25 89‡ 27‡

.7 � 2.5 (4–13) 4.3 � 0.8 (4–7) 4.3

.9 � 3.8 (10–26) 21.1 � 3.5 (10–26) 21.3

complete tumor necrosis.ups.oups.parisons are not significant.

dent t test; Table 2). The time of RF

ablation in each patient was shortestin the RF 3000 group (16.6 minutes �7.5) than in the other three groups(31.7 min � 13.2 in RF 2000 group,28.3 min � 12.0 in RITA group, 27.1min � 12.0 in internally cooled elec-trode group; all P � .005, two-tailedStudent t test; Table 2).

Interactive algorithms were also re-quired more commonly in the RF 2000group than in the RF 3000 group (25 of34 tumors in the RF 2000 group vseight of 35 tumors in the RF 3000; P �.05, �2 test). However, one tumor inthe depth of segment VII in the RF2000 group, two in the RF 2000 group,and one each in the RF 3000, RITA,and internally cooled electrode groupsclose to the hepatic vein larger than 3mm in diameter did not show com-plete necrosis. Hospitalization of pa-tients was similar among the fourgroups (Table 2).

Sixty-one of the 76 patients (80%)exhibited increased baseline �-feto-protein levels that were reduced tonormal (�20 ng/mL) within 3 monthsafter RF ablation. For the 24 patientswith normal baseline �-fetoproteinlevels, the levels remained normal in22 patients (91.7%) at 6 months afterRF ablation. Thirty-three of 61 patients(54%) with normalization of �-fetopro-tein level after ablation experienced arecurrent increase of �-fetoprotein lev-els when recurrence of HCC was de-

ITA Internally Cooled Electrode

24 2429 31

22 222 2

0.3 (1–2) 1.1 � 0.3 (1–2)

12.0 27.1 � 12.0

9 –22 –1.1 (4–8) 4.3 � 1.0 (4–8)

2.6 (11–26) 21.5 � 2.5 (11–26)

le

R

1 �

31 �

5 �

20 �

utgroe gr

tected.

Lin et al • 1123Volume 18 Number 9

Local Tumor Progression and NewHCC Recurrence

After a median of 22 months of fol-low-up (mean, 20 months � 5; range,10–26 months), local tumor recurredin three cases in the RF 2000 group,two in the RF 3000 group, two in theinternally cooled electrode group, andtwo in the RITA group. The cumula-tive local tumor progression rates forthe main tumor at the end of 2 yearswere 12% in the RF 2000 group, 8% inthe RF 3000 group, 8.3% in the inter-nally cooled electrode group, and 8.2%in the RITA group. No significant dif-ference was observed among the fourgroups (P � .37, log-rank test).

New HCC occurred in six patients(24%) in the RF 2000 group, eight pa-tients (32%) in the RF 3000 group,seven patients (28%) in the internallycooled electrode group, and eight pa-tients (32%) in the RITA group. Thecumulative new HCC recurrence ratesat the end of 1 and 2 years were 21%and 35% in the RF 2000 group, 23%and 36% in the RF 3000 group, 22%

and 35% in the internally cooled elec-

trode group, and 22% and 36% in theRITA group. No significant differencewas found among the four groups.

Overall and Cancer-free Survival

During the follow-up period, deathoccurred in three cases (12%) in the RF2000 group, two cases (8%) in the RF3000 group, two cases (8%) in the in-ternally cooled electrode group, andone patient (4%) in the RITA group. Intotal, five patients died of hepatic fail-ure and three died of tumor progres-sion. No significant difference was ob-served among the four groups. Theoverall survival rates at 1 and 2 yearswere 87% and 73% in the RF 2000group, 88% and 75% in the RF 3000group, 90% and 78% in the internallycooled electrode group, and 89% and76% in the RITA group. The cancer-free survival rates at 1 and 2 yearswere 77% and 55% in the RF 2000group, 80% and 56% in the RF 3000group, 79% and 54% in the internallycooled electrode group, and 79% and55% in the RITA group. No significant

Figure. Images frotumor near the brawith RF ablation. (ashowed an enhancedays after RF ablatance in the arterialThree months afterance persisted (arrocomplete ablation.

difference in overall and cancer-free

survival rates was observed amongthe four groups.

Adverse Effects

Although all patients were sedatedbefore RF ablation, severe pain wasexperienced in four patients treatedwith LeVeen electrodes (three in theRF 3000 group and one in the RF 2000group) during RF ablation. The rate ofsevere pain during the RF procedurewas slightly higher in patients treatedwith the LeVeen electrode comparedwith the internally cooled electrodeand the RITA electrode (four of 50 vszero of 50, P � .058, Fisher’s exacttest). The severe pain in these patientswas controlled by higher doses of me-peridine (�1 mg/kg) and midazolaminjection (�0.1 mg/kg) and the proce-dure was completed. No patient hadcholangitis, bile duct injury, or needletract seeding. Transient pleural effu-sion was encountered in two patientsafter RF ablation, and both subse-quently recovered. No other severe

a 67-year-old woman with a 3.7-cm HCCof the right portal vein that was treatedynamic CT imaging before RF ablationumor (arrow) in arterial phase. (b) Two, the tumor showed a hypodense appear-ase (arrow, branch of portal vein). (c)

ablation, the tumor’s hypodense appear-in arterial phase of CT, which represents

mnch) Dd t

ionphRFw)

adverse effect was observed, nor did a

1124 • Comparison of Four RF Devices for HCC Ablation September 2007 JVIR

prominent hemoperitoneum requireblood transfusion in any patient.

Mild fever without chills occurredin nine patients after RF ablation. Nospiking high fever occurred and noliver abscess formed after therapy.Nevertheless, the wound was scarred7–10 days after RF ablation in threecases, with pre–RF ablation asciteswith the LeVeen electrode in the RF2000 and RF 3000 groups.

DISCUSSION

The four RF generators used in ourstudy have also been used for thetreatment of HCC and hepatic metas-tases worldwide. To our knowledge,no prospective study has comparedthese four RF devices, and our resultsare the first that not only showed thatthe RF 3000 requires shorter ablationtime than the other three RF devicesand a less interactive algorithm thanthe RF 2000 device. We have alsoshown similar effectiveness in terms ofcomplete tumor necrosis and local tu-mor progression among the four RFgenerators. The similar effectivenessamong the four RF devices shown inour results was consistent with thosefound in two human prospective stud-ies that compared two RF devices(8,14). In addition, our results alsoshowed greater local tumor progres-sion in tumors larger than 3 cm com-pared with tumors 3 cm or smaller,and this finding was also consistentwith those of previous studies (2,6).

Comparing these four RF devices,RF 2000 and RF 3000 with expandableelectrodes used rapidly increasing im-pedance and then power rolloff as theendpoint of the ablation procedure;therefore, achievement of this end-point could accurately indicate com-plete necrosis (4,6–8,19). The formerdevice had a maximum power of 100W and the latter device had a maxi-mum power of 200 W. The treatmentalgorithms were similar in these twodevices. However, the higher power ofthe RF 3000 generator could reducethe necessity of the interactive algo-rithm and therefore reduce the abla-tion time and slightly increase the rateof complete necrosis, as reported inour results.

The higher power of the RF 3000generator could also achieve completenecrosis in one of two tumors close to

a vein exceeding 3 mm. The RF 2000

generator failed to achieve completenecrosis in two tumors with similarlocations. The same benefit was alsoobserved with the RITA RF devicewith a maximum power of 150 W. Oneof two tumors close to a vein exceed-ing 3 mm could also be completelyablated with the RITA electrode.

The interactive algorithm was re-quired in all tumors close to a veinexceeding 3 mm in our series as a re-sult of a heat-sink effect. With the in-teractive algorithm, the heat could beconcentrated in a smaller zone whenthe expanded electrode was partiallywithdrawn and then redistributed intothe outer zone when the electrode wasfully redeployed after the smaller cen-tral zone had been completely ablated(15,19,20). The limitation of the inter-nally cooled single-electrode devicefor a tumor close to a vein exceeding 3mm was also observed in one case inour series and in other reports (21–23).It is likely that blocking the heat-sinkeffect by inserting a balloon into thevessel immediately before RF ablationis an alternative approach for this con-dition (21,22).

The frequency of severe pain wasslightly higher in patients who re-ceived a larger-gauge (14 gauge)LeVeen electrode (P � .058) versus asmaller-gauge (17 gauge) internallycooled electrode or a 15-gauge RITAelectrode. The major complication ofintraperitoneal hemorrhage was alsomore common with the RF 2000 sys-tem than with the internally cooledelectrode (8). Moreover, our study alsoshowed that wound healing was de-layed in three cases with HCC andascites treated with the larger-gaugeLeVeen electrode compared with sim-ilar cases treated with the internallycooled electrode and RITA electrode.

The limitation of the present studymight be a small sample size thatcould contribute to the nonsignificantdifference in complete tumor necrosisand local tumor progression amongthe four RF devices. However, our re-sults have shown a benefit of theshorter ablation time of the RF 3000system, which is a newer RF devicethan the RF 2000 device, comparedwith the other three RF devices. It isprobably unsuitable or impractical torecruit more patients by using the oldRF device to obtain a significant dif-ference in one of the endpoints. How-

ever, recruitment of more patients

may increase the power of the study.Moreover, the sample size in the stud-ies of Komorizono et al (8) and Shibataet al (14) were also small (38 and 36cases in the Cool-tip and RF 2000groups, respectively), and their resultsalso showed no difference in clinicaleffects between groups.

In conclusion, our results show thatthe RF 3000 system requires shorterablation time than the other three RFdevices and requires a less interactivealgorithm than the RF 2000 device.However, the internally cooled elec-trode and three types of expandableelectrodes yield similar effectivenessin terms of complete tumor necrosisand local tumor progression.

References1. Livraghi T, Goldberg SN, Lazzaroni S,

Meloni F, Solbiati L, Gazelle GS.Small hepatocellular carcinoma: treat-ment with radio-frequency ablationversus ethanol injection. Radiology1999; 210:655–661.

2. Lencioni RA, Aligaier HP, Cioni D, etal. Small hepatocellular carcinoma incirrhosis: randomized comparison ofradiofrequency thermal ablation ver-sus percutaneous ethanol injection. Ra-diology 2003; 228:235–240.

3. McGahan JP, Dodd GD. Radiofre-quency ablation of the liver: currentstatus. AJR Am J Roentgenol 2001; 176:3–16.

4. Curley SA, Izzo F, Delrio P, et al.Radiofrequency ablation of unresect-able primary and metastatic hepaticmalignancies: results in 123 patients.Ann Surg 1999; 230:1–8.

5. Omata M, Tateishi R, Yoshida H, Shi-ina S. Treatment of hepatocellularcarcinoma by percutaneous tumor ab-lation methods: ethanol injection ther-apy and radiofrequency ablation. Gas-troenterology 2004; 127(suppl):S159–S166.

6. Lin SM, Lin CJ, Lin CC, Hsu CY,Chen YC. Radiofrequency ablationimproves prognosis compared withethanol injection for hepatocellular car-cinoma �4 cm. Gastroenterology 2004;127:1714–1723.

7. Morimoto M, Sugimori K, Shirato K, etal. Treatment of hepatocellular carci-noma with radiofrequency ablation: ra-diologic-histologic correlation duringfollow-up periods. Hepatology 2002; 35:467–475.

8. Komorizono Y, Oketani M, Sako K, etal. Risk factors for local recurrence ofsmall hepatocellular carcinoma tumors

after a single session, single application

Lin et al • 1125Volume 18 Number 9

of percutaneous radiofrequency abla-tion. Cancer 2003; 97:1253–1262.

9. Mazzaferro V, Battiston C, PerroneS, et al. Radiofrequency ablation ofsmall hepatocellular carcinoma in cir-rhotic patients awaiting liver trans-plantation: a prospective study. AnnSurg 2004; 240:900–909.

10. Livraghi T, Solbiati L, Meloni MF et al.Treatment of focal liver tumors withpercutaneous radio-frequency abla-tion: complications encountered in amulticenter study. Radiology 2003; 226:441–451.

11. Llovet JM, Vilana R, Bro C, et al.Increased risk of tumor seeding afterpercutaneous radiofrequency ablationfor single hepatocellular carcinoma.Hepatology 2001; 33:1124–1129.

12. Shiina S, Teratani T, Obi S, et al. Arandomized controlled trial of radio-frequency ablation with ethanol injec-tion for small hepatocellular carci-noma. Gastroenterology 2005; 129:122–310.

13. Lencioni R, Llovet JM. Percutaneousethanol injection for hepatocellular car-cinoma: alive or dead. J Hepatol 2005;43:377–380.

14. Shibata T, Shibata T, Maetani Y, Isoda H,

Hiraoka M. Radiofrequency ablation

for small hepatocellular carcinoma:prospective comparison of internallycooled electrode and expandable elec-trode. Radiology 2006; 238:346–353.

15. Lin SM, Lin CJ, Chung HJ, Hsu CW,Peng CY. Power rolloff during inter-active radiofrequency ablation canenhance necrosis when treating he-patocellular carcinoma. AJR Am JRoentgenol 2003; 180:151–157.

16. Meloni MF, Goldberg SN, LivraghiT, et al. Hepatocellular carcinomatreated with radiofrequency ablation:comparison of pulse inverse contrast-en-hanced harmonic sonography, contrast-enhanced power Doppler sonographyand helical CT. AJR Am J Roentgenol2001; 177:375–380.

17. Goldberg SN, Stein MC, Gazelle GS,Sheiman RG, Kruskal JB, Clouse ME.Percutaneous radiofrequency tissueablation: optimization of pulsed-radio-frequency technique to increase coagu-lation necrosis. J Vasc Interv Radiol1999; 10:907–916.

18. Goldberg SN, Grassi CJ, Cardella JF,et al. Image-guided tumor ablation:standardization of terminology and re-porting criteria. J Vasc Interv Radiol

2005; 235:728–739.

19. Arata MA, Nisenbaum HL, Clark TWI,Soulen MC. Percutaneous radiofre-quency ablation of liver tumors withthe LeVeen probe: is power roll-offpredictive of response. J Vasc IntervRadiol 2001; 12:455–458.

20. Kobayshi M, Ikeda K, Someya T, et al.Stepwise hook extension technique forradiofrequency ablation therapy ofhepatocellular carcinoma. Oncology2002; 63:139–144.

21. de Baere T, Bessoud B, Dromain C, etal. Percutaneous radiofrequency ab-lation of hepatic tumors during tempo-rary venous occlusion. AJR Am JRoentgenol 2002; 178:53–59.

22. Yamasaki T, Kurokawa F, ShirahashiH, Kusano N, Hironaka K, Okita K.Percutaneous radiofrequency ablationtherapy for patients with hepatocellu-lar carcinoma during occlusion of he-patic blood flow: comparison withstandard percutaneous radiofrequencyablation therapy. Cancer 2002; 95:2353–2360.

23. Lu DSK, Raman SS, Vodopich DJ,Wang M, Sayre J, Lassman C. Effectof vessel size on creation of hepaticradiofrequency lesions in pigs: assess-ment of the “heat sink” effect. AJR

Am J Roentgenol 2002; 178:47–51.