Efficacy of Selective Transarterial Chemoembolizationin Inducing Tumor Necrosis in Small (<5 cm)

Hepatocellular CarcinomasRita Golfieri,1 Alberta Cappelli,1 Alessandro Cucchetti,2 Fabio Piscaglia,3 Maria Carpenzano,1 Eugenia Peri,2

Matteo Ravaioli,2 Antonia D’Errico-Grigioni,4 Antonio Daniele Pinna,2 and Luigi Bolondi3

Transarterial chemoembolization (TACE) is commonly used as a bridge therapy for patientsawaiting liver transplantation (LT) and for downstaging patients initially not meeting the Mi-lan criteria. The primary aim of this study was to analyze whether a difference exists betweenselective/superselective and lobar TACE in determining tumor necrosis by a pathological analy-sis of the whole lesion at the time of LT. The secondary aim was to investigate the relationshipbetween the tumor size and the capacity of TACE to induce necrosis. Data were extracted froma prospective database of 67 consecutive patients who underwent LT for hepatocellular carci-noma and cirrhosis from 2003 to 2009 and were treated exclusively with TACE as a bridging(n 5 53) or downstaging therapy (n 5 14). We identified 122 nodules; 53.3% were treatedwith selective/superselective TACE. The mean histological necrosis level was 64.7%; completetumor necrosis was obtained in 42.6% of the nodules. In comparison with lobar TACE, selec-tive/superselective TACE led to significantly higher mean levels of necrosis (75.1% versus52.8%, P 5 0.002) and a higher rate of complete necrosis (53.8% versus 29.8%, P 5 0.013).A significant direct relationship was observed between the tumor diameter and the mean tumornecrosis level (59.6% for lesions < 2 cm, 68.4% for lesions of 2.1-3 cm, and 76.2% for lesions> 3 cm). Histological necrosis was maximal for tumors > 3 cm: 91.8% after selective/superse-lective TACE and 66.5% after lobar procedures. Independent predictors of complete tumor ne-crosis were selective/superselective TACE (P 5 0.049) and the treatment of single nodules (P5 0.008). Repeat sessions were more frequently needed for nodules treated with lobar TACE(31.6% versus 59.3%, P 5 0.049). Conclusion: Selective/superselective TACE was more suc-cessful than lobar procedures in achieving complete histological necrosis, and TACE was moreeffective in 3- to 5-cm tumors than in smaller ones. (HEPATOLOGY 2011; 53:1580-1589)

Transarterial chemoembolization (TACE) is therecommended treatment strategy for patientswith intermediate-stage hepatocellular carci-

noma (HCC) according to the American Associationfor the Study of Liver Diseases guidelines.1,2 In thesetting of liver transplantation (LT), TACE is appliedboth to reduce the dropout rate for patients on thewaiting list (bridge therapy) and to downstage patientswith HCC not initially meeting the transplantabilitycriteria (downstaging protocols).3

The capability of TACE to induce extensive tumornecrosis is still under debate, and this technique isconsidered to be a noncurative modality. Whether thisbelief derives from the real potential of the techniqueor from the fact that it has mainly been applied totumors that are large and are, consequently, more diffi-cult to treat is still a matter of discussion. Similarly,the role of the various technical modalities of TACEprocedures in determining the final rate of necrosis hasnot been adequately investigated in Western countries.The recommendation for TACE as the standard of

care for intermediate-stage HCC is based on the

Abbreviations: CEUS, contrast-enhanced ultrasonography; CT, computedtomography; HCC, hepatocellular carcinoma; LT, liver transplantation; MC,Milan criteria; MRI, magnetic resonance imaging; PEI, percutaneous ethanolinjection; TACE, transarterial chemoembolization.From the 1Radiology Unit, 2Liver and Multiorgan Transplant Unit, 3Division

of Internal Medicine, Department of Digestive Diseases and Internal Medicine,and 4Pathology Division, Felice Addarii Institute, Department of Oncology andHematology, Sant’Orsola-Malpighi Hospital, University of Bologna, Bologna,Italy.Received October 1, 2010; accepted February 3, 2011.Address reprint requests to: Alberta Cappelli, M.D., Unita Operativa

Radiologia Malpighi, Azienda Ospedaliero-Universitaria di Bologna,Policlinico Sant’Orsola-Malpighi, Via Albertoni 15, 40138 Bologna, Italy.E-mail: alberta.cappelli@aosp.bo.it; fax: þ39-051-6362699.CopyrightVC 2011 by the American Association for the Study of Liver Diseases.View this article online at wileyonlinelibrary.com.DOI 10.1002/hep.24246Potential conflict of interest: Nothing to report.Additional Supporting Information may be found in the online version of

this article.

1580

demonstration of improved survival in comparisonwith the best supportive care or suboptimal therapiesin a meta-analysis of six randomized control trials.4

However, there was considerable heterogeneity betweenthe individual study designs of the six trials, and thedifferences included the patient populations andTACE techniques. More specifically, the oldest trials ofthe meta-analysis included lobar or whole liver emboli-zation (i.e., the injection of a mixture of Lipiodol, achemotherapeutic agent, and an embolizing materialinto either the main lobar artery or the hepatic arteryitself ), whereas more recently, selective treatments havebeen used (i.e., the injection of agents into the seg-mental or subsegmental branches feeding the tumors)with apparently better survival results.5,6 On the basisof radiological features, selective/superselective TACEhas already been shown to be capable of obtaining ahigher rate of tumor necrosis than conventional lobaror whole liver TACE.7 However, whether the findingsof improved survival with selective techniques reallycorrespond to an improved necrotizing capability,reduced liver toxicity, or both has never been eluci-dated on the basis of histological findings in a suffi-ciently large Western population.The results of studies published in the Asiatic litera-

ture suggest that segmental or subsegmental TACE hasbeen more effective and has resulted in higher rates oftumor necrosis (64%-83%) than proximal/whole liverTACE (approximately 38%) in historical series.8-11

Even though the efficacy of TACE can be reliablyassessed only by the measurement of tumor necrosisduring a histological examination of the whole tumor,only three of these series8,10,11 included surgicallyremoved nodules, and the histological quantification ofnecrosis involved small sample sizes (11, 12, and 7lesions, respectively).However, in the Western literature, the advantages

of selective embolization have not been well reportedbecause nonselective TACE has been performed evenin recent studies.12

Therefore, the primary aim of this study was to ana-lyze whether a difference exists between selective/super-selective and lobar TACE in determining tumor necro-sis by a pathological analysis of the whole lesion at thetime of transplantation. The secondary aim was toinvestigate the relationship between the tumor size andthe capacity of TACE to induce necrosis.

Patients and Methods

Data were extracted from a prospectively collecteddatabase for 118 consecutive patients who had a pre-

transplant diagnosis of HCC resulting from cirrhosis,underwent LT between January 1, 2003 and December31, 2009 at the Liver and Multiorgan Transplant Unitof Sant’Orsola-Malpighi Hospital, and were treatedwith bridging or downstaging procedures. The finalstudy population consisted of 67 patients treated onlywith TACE (performed exclusively at our tertiary careinstitution), as outlined in Fig. 1 and Table 1, with 53patients meeting the Milan criteria (MC) and 14 meet-ing our downstaging protocol.3,13 Before undergoingTACE, all patients were assessed (1) to define the degreeof liver function by laboratory examinations and (2) todetect and characterize all liver nodules by imaging tech-niques. The Child-Pugh score and the Model for End-Stage Liver Disease score (the latter according to the for-mula proposed by Freeman et al.14) were calculated.The patients were staged according to the United Net-work for Organ Sharing guidelines15 and the integratedBarcelona Clinic Liver Cancer staging system.16

Our downstaging protocol is based on radiologicalfindings at the initial assessment of (1) a single HCC� 8 cm, (2) bifocal HCCs < 5 cm, or (3) up to fivenodules (each � 4 cm) with a total tumor diameter �12 cm, as previously reported.4,13 The last patientswere listed for LT once their HCCs were successfullydownstaged to meet the MC. The criteria for success-ful downstaging were based at that time only on themaximum diameter of tumors with imaging signs ofvital tissue, whatever its extent within the tumorswas.1,2,17 Exclusion criteria from the waiting listincluded evidence of gross vascular invasion, tumorprogression beyond the limits of the MC, and evi-dence of extrahepatic or lymph node metastases. Portalthrombosis was not an exclusion criterion if it couldbe shown to be nonneoplastic.18

Imaging Evaluation. Since 2003 (when the studybegan), our technical requirements for contrast-enhanced computed tomography (CT) and magneticresonance imaging (MRI) have met the minimal crite-ria subsequently recommended by the American con-sensus on the diagnostic assessment of liver nodules inpatients on the waiting list for LT.19 For CT, four con-trast phases were carried out after precontrast scans(early and late arterial, venous, and late), whereas onlythree phases were carried out for MRI (arterial, ve-nous, and late). The diagnosis was established accord-ing to the latest international guidelines on the man-agement of HCC (i.e., the European Association forthe Study of the Liver guidelines from 200117 and theAmerican Association for the Study of Liver Diseasesguidelines from 20051,2). Whenever needed, CT orMRI was used along with low–mechanical index

HEPATOLOGY, Vol. 53, No. 5, 2011 GOLFIERI ET AL. 1581

contrast-enhanced ultrasonography (CEUS) with Sono-vue (Bracco, Milan, Italy).Since 2006, all studies have been evaluated with the

support of the institutional picture archiving and com-munication system (Carestream, version 1.1, Kodak,Rochester, NY), and the radiological reports stored inthe radiology information system (e-ris, Exprivia Pro-ject SpA, Rome, Italy) were used for this study. Beforethen, the images had instead been printed on the filmsused by radiologist to make their reports.TACE Protocol. Two different techniques were

applied to treat HCC nodules: lobar and selective/superselective. With the selective/superselective tech-nique, the tumor-feeding arteries were catheterizedwith a highly flexible coaxial microcatheter (a 2.7- to2.8-Fr Terumo Progreat microcatheter or a Boston Sci-entific Renegade HI-FLO microcatheter) passedthrough a 4-Fr catheter previously placed approxi-mately in the hepatic artery itself. More specifically, forselective TACE, the tip of the microcatheter wasplaced into the hepatic arterial branch afferent to thesegment in which the tumor was located. In the caseof superselective TACE, the tip of the catheter was fur-ther advanced into the subsegmental branches feedingthe tumor (Fig. 2A,B).After microcatheter placement, a mixture of epirubi-

cin (Pfizer) and Lipiodol (Guerbet) with an average

total volume of 50 mL (range ¼ 40-75 mL) wasinjected under fluoroscopic control, and this was fol-lowed by embolization with Spongel (GelitaSpon gel)particles until there was complete stasis in the tumor-feeding vessels. In all cases, there was testing for a par-asitic tumor blood supply through accessory arteries(i.e., the inferior phrenic, internal mammary, or inter-costal arteries; Fig. 3), and if one was present, thepatient underwent additional superselective treatment(a chemotherapeutic mixture plus embolization).Nonselective lobar TACE consisted of the injection

of the same treatment material used in the selectiveprocedures into the right or left lobar branches andthen embolization (Fig. 4). Consequently, a largerregion (usually the whole lobe containing the tumors)was treated.A selective or, if possible, superselective TACE pro-

cedure was the preferred modality whenever it wastechnically feasible. In all other cases (i.e., when there

Fig. 1. Flow chart of the study population. Abbreviation: PEI, percu-taneous ethanol injection.

Table 1. Baseline Clinical and Tumor Characteristics of theStudy Population

Variable Value

Age at start (years) 55.5 6 7.9/57 (37-67)

Male gender 60 (89.6%)

Serum creatinine (mg/dL) 0.97 6 0.22/0.90 (0.60-1.87)

Serum bilirubin (mg/dL) 2.36 6 2.21/1.85 (0.88-14.67)

International normalized ratio 1.38 6 0.20/1.35 (1.06-2.28)

Child-Pugh class A 54 (80.6%)

Model for End-Stage Liver Disease score 12.9 6 3.6/12.5 (7-26)

Hepatitis B positive serology 16 (23.9%)

Hepatitis C positive serology 46 (68.7%)

United Network for Organ Sharing

T stage at start

T1 10 (14.9%)

T2 43 (64.2%)

T3-T4a 14 (20.9%)

Barcelona Clinic Liver Cancer stage at start

A (within MC) 53 (79.1%)

B (outside MC) 14 (20.9%)

Single nodule 32 (47.8%)

Number of nodules 2 (1-4)

Diameter of largest tumor (cm) 2.7 6 0.9/2.7 (1.2-5.0)

Sum of tumor diameters (cm) 4.0 6 1.8/3.6 (1.2-8.4)

Serum alpha-fetoprotein (ng/mL) 14 (2-735)

Selective/superselective TACE 38 (56.7%)

Lobar TACE 27 (40.3%)

Combination of selective and lobar TACE 2 (3.0%)

Repeat TACE 29 (43.3%)

There were 67 patients in this study. Continuous variables are reported as

means and standard deviations, medians and ranges, or both; other variables

are reported as numbers and percentages. The MC included United Network for

Organ Sharing stages T1 and T2. The tumors were classified as T3 to T4a and

fulfilled previously published downstaging eligibility criteria.3,13 The only Barce-

lona Clinic Liver Cancer stages were A and B (which corresponded to those ful-

filling the MC and those not fulfilling the MC, the latter being intermediate HCC

patients meeting our downstaging criteria) because no patient with portal or he-

patic vein infiltration or with distant metastasis is eligible for transplantation at

our center.

1582 GOLFIERI ET AL. HEPATOLOGY, May 2011

was multinodular disease in one lobe with a nodule ornodules fed by multiple arteries or when the catheteri-zation of the tiny tumor-feeding vessels was not possi-ble), lobar TACE was performed.A CT scan was performed approximately 30 days af-

ter the procedure to detect both Lipiodol retentionwithin the tumor and residual viable tumor tissue. Ho-mogeneous and dense Lipiodol uptake with no addi-tional contrast enhancement was considered to corre-spond to a complete response. When this was not thecase and residual viable tumors were confirmed bycomplementary imaging studies (MRI or CEUS) ornew lesions had developed but the patients maintainedadequate hepatic function and reserve, they werereferred for repeat procedures. TACE treatment wasrepeated on demand, that is, in patients with residualor recurrent tumors observed by CT or MRI, accord-ing to the amended Response Evaluation Criteria inSolid Tumors guidelines and in agreement with recentexpert opinion.20

The CT or MRI protocol after a TACE procedurewas the same as that applied before TACE. A viabletumor was defined by contrast agent uptake in the ar-terial phase and washout in the portal phase and/orlate phase. During the CT scan, contrast enhancementwas visually assessed by a comparison of the unen-hanced and arterial phase images to distinguishbetween iodized oil and contrast agent enhancement.In doubtful cases, CEUS, MRI, or both were per-formed, as previously described.Histopathology. After LT, all the livers were exam-

ined by two experienced hepatobiliary pathologists.The livers were sectioned into 10-mm-thick sections.All identified nodules were grossly described with

respect to the site, size, types of margins (vaguely/dis-tinctly nodular or infiltrative), and necrosis, and theywere completely paraffin-embedded. Multiple 3-lmsections were stained with hematoxylin and eosin, re-ticulin, periodic acid Schiff with and without diastase,and Perls iron stain. The percentage of tumor necrosiswas calculated as the proportion of the necrotic tissuewith respect to the total tumor area and was based onthe expert judgment of the pathologist in 10% steps.A necrosis rate of 100% was assumed to indicate com-plete necrosis; a rate of 99% or less was considered toindicate incomplete necrosis. For the purpose of

Fig. 2. Hepatic angiographic examination. Left: The cranial portion of this large (>3-cm) HCC of the right hepatic lobe was previously treated(the quite visible Lipiodol accumulation is indicated by the dashed arrow). One month after the first TACE procedure, a residual viable tumorwas evident in the peripheral portion of the nodule on a CT scan, and additional TACE treatment was required. A superselective contrast injectioninto the distal subsegmental branches of the right hepatic artery (the tip of the microcatheter is indicated by the black arrow) showed two smallvessels feeding the viable caudal portion (indicated by the empty arrow), which were then subjected to TACE. Right: A postprocedural controlshowed the complete devascularization of the target nodule with persistent flow into the proximal trunk of the catheterized vessel.

Fig. 3. Angiographic examination. Selective TACE was performedafter catheterization of the proximal trunk of the right inferior phrenicartery (the microcatheter is indicated by the black arrow); this contrib-uted to the perfusion of the HCC (the tumor stain is indicated by theempty arrow) located in the dome of the liver (segment VII; the domeis indicated by the dashed arrow).

HEPATOLOGY, Vol. 53, No. 5, 2011 GOLFIERI ET AL. 1583

evaluating the percentage of complete necrosis accord-ing to tumor size, the HCCs were grouped by size:�2, 2.1 to 3.0, and 3.1 to 5.0 cm.Statistical Analysis. Continuous variables were

reported as means and standard deviations, medians andranges, or both; the differences between the subgroupswere analyzed with the t test after the Levene correction,analysis of variance, or the Mann-Whitney test as appro-priate. Categorical variables were reported as numbersand percentages, and the differences between the sub-groups were analyzed with the chi-square test with aYates correction. The amounts of tumor necrosis werereported both as continuous variables and as semiquanti-tative values, and the differences between subgroupswere calculated. In order to identify the potential rela-tionships between the covariates with respect to tumornecrosis, all variables significantly affecting tumor necro-sis in the univariate analysis were entered into a multi-variate logistic regression model to identify the inde-pendent predictors of complete tumor necrosis. A Pvalue < 0.05 was considered statistically significant inall cases. Statistical analysis was carried out with SPSS13.0 (SPSS, Inc., Chicago, IL).

Results

The baseline clinical and tumor characteristics ofthe study group are reported in Table 1. Thirty-eightof the 67 patients underwent selective/superselectiveTACE exclusively (56.7%), 27 patients underwent lo-bar TACE exclusively (40.3%), and 2 patients weretreated with a combination of the two techniques(3%). In the latter two cases, lobar TACE and selectiveTACE were each used in only one lobe, and thisallowed an assessment of the treatment outcome foreach technique.In order to limit the risk of liver decompensation,

we never performed whole lobe treatments of bothlobes in the same session (or whole liver treatments).

Thirty-eight patients had a single course of TACE(56.7%), and the remaining 29 had two or morecourses (43.3%). The median time from the firstTACE procedure to LT was 8.7 months (range ¼ 1-32months), and the median time on the waiting list was6.2 months (range ¼ 1-29 months). For patients whounderwent more than one TACE session, the mediantime from the last imaging procedure to LT was 2.6months (range ¼ 1-92 days).No patient of the present series experienced major

complications related to the procedure, and noneunderwent LT within 30 days of the procedure (thiscould be interpreted as an expression of rapid deterio-ration of liver function). The median hospital stayswere 4.5 days after lobar procedures (range ¼ 2-65days) and 3.5 days after selective/superselective TACE(range ¼ 2-56 days, P ¼ 0.651); clinically significantfever (maximum temperature > 38�C) occurred in 20cases after lobar TACE (74.0%) and in 23 cases afterthe selective procedure (60.5%, P ¼ 0.255). Detailsregarding the course of liver laboratory tests afterTACE are reported in Table 2.Clinical and tumor characteristics with respect to lo-

bar and selective/superselective TACE are reported inTable 3: as expected, the choice of the procedure wasaffected by the presence of multinodular tumors, butit was unaffected by liver function status, although aminimal trend toward worse liver function in thosetreated with selective/superselective TACE emerged.Patients who underwent selective/superselective TACErequired fewer repeat procedures than patients whounderwent lobar TACE [12 of 38 (31.6%) versus 16of 27 (59.3%), P ¼ 0.0049] because residual vitaltumors were less common. One of the two patientswho received the combination of techniques requiredone additional treatment.Histological Necrosis According to the TACE

Modality. Because the type of TACE performed ineach patient was affected by the number of tumors

Fig. 4. Angiographic examination before(left) and immediately after lobar TACE (right).Left: The treatment mixture was injected intothe right hepatic artery (the distal part of thecatheter in the right hepatic artery is indi-cated by the black arrow), which also fed theHCC in segment VI (the tumor stain is indi-cated by the empty arrow). Right: A postpro-cedural control demonstrated the completeuptake of the Lipiodol-drug emulsion and thedisappearance of any perfusion within the tu-mor along with the absence of perfusion inthe peripheral arterial branches of almost thewhole right lobe.

1584 GOLFIERI ET AL. HEPATOLOGY, May 2011

and the stage, an analysis of the histological outcomewas carried out for each individual nodule. At the be-ginning of the observation period, 122 nodules wereidentified; 53.3% (65 cases) were treated with selec-tive/superselective TACE, whereas the remaining46.7% were treated with a lobar procedure (57 cases).The characteristics of the treated nodules with respectto the adopted procedures are reported in Table 4; thediameters of the nodules treated with selective/superse-lective and lobar TACE were similar (P ¼ 0.725), butas expected, multiple tumors were more frequentlytreated with lobar TACE (P ¼ 0.041). In theexplanted liver, the mean treated tumor necrosis levelwas 64.7%; complete tumor necrosis was obtained in

42.6% (52 cases), whereas the remaining proportionshowed different degrees of necrosis.Tumor necrosis was affected by the adopted proce-

dure; it was greater after selective/superselective TACE(75.1%) versus lobar TACE (52.8%, P ¼ 0.002)whether all the nodules were considered as a whole orthe nodules were subgrouped according to their size(Table 4). Complete necrosis and necrosis � 90%were more frequently observed after selective/superse-lective TACE versus lobar TACE (P ¼ 0.013 and P ¼0.008, respectively).The treatment of patients with single nodules led to

higher levels of tumor necrosis (mean ¼ 86.1%) thanthe treatment of patients with multiple nodules

Table 2. Clinical Course of Liver Function Parameters After TACE With Respect to the Adopted Procedure

Variable Baseline Value (n 5 67) Day 1 (n 5 67) Day 3 (n 5 60) Day 5 (n 5 39)

Serum creatinine (mg/dL) 0.90 (0.60-1.87) 0.96 (0.65-1.79) 0.94 (0.60-1.67) 0.93 (0.48-1.91)

Serum bilirubin (mg/dL) 1.85 (0.88-14.67) 2.44 (0.92-16.61) 2.95 (0.89-16.11) 2.66 (1.16-16.93)

International normalized ratio 1.35 (1.06-2.28) 1.39 (0.84-2.11) 1.40 (0.89-2.46) 1.40 (1.16-2.10)

Model for End-Stage Liver Disease score 12.5 (7-26) 14 (9-26) 14 (9-27) 14 (9-27)

Lobar TACE (n) 27 27 25 14

Serum creatinine (mg/dL) 0.94 (0.60-1.50) 0.96 (0.70-1.70) 0.96 (0.74-1.45) 0.97 (0.70-1.35)

Serum bilirubin (mg/dL) 1.59 (0.88-4.64) 2.12 (0.9-5.20) 2.84 (0.89- 7.98) 2.64 (1.16-9.84)

International normalized ratio 1.29 (1.10-1.78) 1.33 (1.21-1.70) 1.34 (1.19-1.80) 1.39 (1.18-1.68)

Model for End-Stage Liver Disease score 12 (8-19) 13 (9-21) 14 (9-21) 14 (10-21)

Selective/superselective TACE (n) 38 38 33 18

Serum creatinine (mg/dL) 0.90 (0.76-1.87) 0.96 (0.6-1.79) 0.90 (0.60-1.67) 0.91 (0.48-1.91)

Serum bilirubin (mg/dL) 2.01 (0.59-14.67) 2.53 (0.92-16.61) 3.01 (1.06-16.11) 2.45 (1.3-16.93)

International normalized ratio 1.37 (1.06-2.28) 1.45 (0.84-2.11) 1.45 (0.89-2.46) 1.43 (1.16-2.10)

Model for End-Stage Liver Disease score 13 (7-26) 14 (10-26) 15 (8-27) 14 (9-27)

Continuous variables are reported as medians and ranges. Data for all patients include two cases with a combination of the two techniques. The data for day 5

pertain to fewer patients because only those not recovering fast enough were still in the hospital. Those doing well had already been discharged. This might have

led to an overestimation of the worsening of liver function.

Table 3. Baseline Clinical and Tumor Characteristics of the Study Population With Respect to the Adopted TACE Procedure

Variable

TACE Procedure

P ValueLobar (n 5 27) Selective/Superselective (n 5 38)

Age at start (years) 57 (43-65) 55.5 (37-67) 0.337

Male gender 25 (92.6%) 35 (92.1%) 0.942

Serum creatinine (mg/dL) 0.94 (0.60-1.50) 0.90 (0.76-1.87) 0.272

Serum bilirubin (mg/dL) 1.59 (0.88-4.64) 2.01 (0.59-14.67) 0.123

International normalized ratio 1.29 (1.10-1.78) 1.37 (1.06-2.28) 0.150

Child-Pugh class A 22 (81.5%) 30 (78.9%) 0.801

Model for End-Stage Liver Disease score 12 (8-19) 13 (7-26) 0.170

Hepatitis B positive serology 7 (25.9%) 9 (23.7%) 0.836

Hepatitis C positive serology 19 (70.4%) 25 (65.8%) 0.697

United Network for Organ Sharing T stage at start 0.073

T1 3 (11.1%) 7 (18.4%)

T2 15 (55.6%) 27 (71.1%)

T3-T4a 9 (33.3%) 4 (10.5%)

Single nodule 10 (37.0%) 22 (57.9%) 0.097

Number of nodules 2 (1-4) 1 (1-4) 0.079

Diameter of largest tumor (cm) 3.0 (1.2-5.0) 2.5 (1.5-4.5) 0.052

Sum of tumor diameters (cm) 5.1 (1.2-8.4) 3.1 (1.5-6.7) 0.016

Serum alpha-fetoprotein (ng/mL) 17 (2-735) 10 (2-382) 0.233

Repeated TACE 16 (59.3%) 12 (31.6%) 0.049

Continuous variables are reported as median and ranges; other variables are reported as numbers and percentages. The two cases treated with a combination

of selective and lobar TACE are not included in this comparison.

HEPATOLOGY, Vol. 53, No. 5, 2011 GOLFIERI ET AL. 1585

(57.1%, P ¼ 0.001). The differences between thetreatment modalities (selective TACE was better thanlobar TACE) were more evident for multiple nodules(P ¼ 0.029; Table 4) than for single nodules (P ¼0.172; Table 4).Histological Necrosis According to the Tumor

Size. A significant direct relationship between necrosisand the tumor diameter was found, regardless of the typeof TACE procedure, in our series of small HCCs: thegreater the tumor diameter, the greater the percentage ofnecrosis. The mean necrosis levels were 59.6% for nodules� 2 cm, 68.4% for nodules of 2.1 to 3.0 cm, and 76.2%for nodules > 3.1 cm (P ¼ 0.038; Table 4).As for an analysis of the relationship between the

achievement of complete tumor necrosis, the TACE mo-dality, and the tumor diameter, the proportion of patientsin each subgroup was too small for a statistically signifi-cant comparison to be made. However, because the attain-ment of complete necrosis resulted from the interaction ofthe aforementioned variables, a multivariate logisticregression analysis was run: in the study population, inde-pendent predictors for achieving complete tumor necrosiswere selective/superselective TACE [Exp(B) ¼ 2.192,95% confidence interval ¼ 1.002-4.793, P ¼ 0.049] andthe treatment of a single nodule [Exp(B) ¼ 3.756, 95%confidence interval ¼ 1.404-10.045, P ¼ 0.008]. Thenodule diameter played a minor role [Exp(B) ¼ 1.656,95% confidence interval¼ 0.926-2.961, P ¼ 0.089].Histological Necrosis and Post-TACE CT Assess-

ment. The post-TACE CT scan showed homogeneousand dense Lipiodol uptake in all nodules in 44 of 67

patients (65.7%) who were considered complete res-ponders. CT results were considered suspicious forincomplete treatment in 5 patients (7.4%) in whomsubsequent CEUS or MRI confirmed viable tumor tis-sue; in the remaining 18 patients (26.9%), at least onenodule showed incomplete Lipidol uptake on a CTscan.The 44 patients with an apparently complete

response were affected by 71 nodules. The 23 patientswith suspicious or incomplete Lipiodol uptake had 51nodules: 24 with complete Lipiodol uptake and 27with incomplete Lipiodol uptake. In 53 (55.8%) ofthe 95 nodules with an apparently complete radiologi-cal response (dense Lipiodol uptake), complete histo-logical necrosis was confirmed. In all 23 patients witha suspicious or incomplete response, a histological ex-amination confirmed vital tissue.

Discussion

Taking advantage of the fact that LT offers the pos-sibility of assessing histological tumor necrosis aftertreatment with TACE, we have been able to show thatthe possibility of performing a selective/superselectiveprocedure is a highly relevant factor in determining tu-mor necrosis.At present, TACE is one of the most widely used

pre-LT treatments in patients with HCCs. The degreeof tumor necrosis induced by TACE has already beenreported in the literature,6,21-29 and there have beendifferent results due to different classifications of the

Table 4. Characteristics of the Nodules in the Explanted Livers With Respect to the Adopted Procedure

Variable All Nodules (122 in 67 Patients)

TACE Procedure

P ValueLobar (57 in 27 Patients) Selective/Superselective (65 in 38 Patients)

Degree of necrosis (%) 64.7 6 40.3 52.8 6 32.0 75.1 6 30.9 0.002

Complete necrosis (100%) 52 (42.6%) 17 (29.8%) 35 (53.8%) 0.013

Necrosis � 90% 65 (53.3%) 23 (40.4%) 43 (66.2%) 0.008

Necrosis < 50% 40 (32.8%) 25 (43.9%) 15 (23.1%) 0.025

Diameter of nodules (cm) 2.3 6 0.9 2.3 6 0.9 2.2 6 1.0 0.725

Number of nodules 0.041

Single 32 (26.2%) 10 (17.5%) 22 (33.8%)

Degree of necrosis (%) 86.1 6 29.3 75.5 6 39.1 90.1 6 23.1 0.172

Multiple 90 (73.8%) 47 (82.5%) 43 (66.2%)

Degree of necrosis (%) 57.1 6 31.0 48.1 6 31.4 67.0 6 28.7 0.029

Nodule diameter class 0.302

�2.0 cm 70 (57.4%) 31 (54.4%) 39 (60.0%)

Degree of necrosis (%) 59.6 6 31.8 50.3 6 31.3 67.0 6 29.2 0.047

2.1-3.0 cm 31 (25.4%) 13 (22.8%) 18 (27.7%)

Degree of necrosis (%) 68.4 6 30.7 45.4 6 41.9 85.0 6 31.4 0.009

3.1-5.0 cm 21 (17.2%) 13 (22.8%) 8 (12.3%)

Degree of necrosis (%) 76.2 6 32.6 66.5 6 38.5 91.8 6 6.5 0.038

No other clinical variables were found to be related to the degree of necrosis on pathological examination; variables significantly related to the degree of necrosis

were included in the multivariate logistic regression model to identify independent predictors of complete tumor necrosis (see the main text). Continuous variables

are reported as means and standard deviations; other variables are reported as numbers and percentages.

1586 GOLFIERI ET AL. HEPATOLOGY, May 2011

tumor necrosis rate, different TACE techniques, and,frequently, small sample sizes. Therefore, the effective-ness of TACE in achieving complete tumor necrosisand, consequently, the proper control of tumor pro-gression still has to be clarified. Theoretically, necrosisresulting from treatment provides a beneficial effect bylimiting the number of dropouts.The present analysis shows that the main determi-

nant in successful treatment is the adopted proceduremodality. In fact, the present data show that the use ofselective/superselective TACE leads to the completenecrosis of HCCs approximately 2 times more oftenthan lobar TACE. To the best of our knowledge, atpresent, only one article30 has reported a differencebetween superselective TACE and lobar TACE andtheir correlation with histological necrosis: the superse-lective procedure was more often associated with com-plete tumor necrosis.Interestingly, we also found that patients treated with

selective/superselective TACE needed to undergo repeatprocedures less frequently. In fact, we repeated TACEonly in cases with persistent vital tumors; selectiveTACE led to a higher rate of complete necrosis andthus limited the need for additional sessions in compari-son with whole liver TACE. This might be beneficial inpreventing progressive liver and vascular damage.However, we should acknowledge that the present

study is not a prospective, randomized study comparingdifferent treatment modalities, and lobar procedureswere performed when selective/superselective ones couldnot be technically carried out for a variety of reasons(but mainly because of the vascular anatomy). It is,therefore, impossible to state with certainty that theresults would still have differed if lobar TACE had beenperformed in patients undergoing selective/superselectiveTACE, who probably had a more favorable vascularanatomy. Nonetheless, according to the data available inthe literature7 and the current study, we recommendpursuing all technical efforts and attempts to carry outselective/superselective TACE. This statement mayappear obvious. However, selective/superselective TACEis usually more time-consuming, more expensive interms of angiography room occupancy and disposablematerials, and more technically demanding than conven-tional TACE; thus, inexperienced or overloaded opera-tors may be tempted to routinely carry out lobar TACE,which should instead be avoided.We also found that the treatment of single nodules

significantly affected tumor necrosis: single nodulesshowed a higher degree of mean tumor necrosis(86.1%) than multiple nodules (57.1%, P ¼ 0.001).Patients with a single nodule who were treated with

selective/superselective or lobar TACE tended to havea higher percentage of necrosis in comparison withpatients with multiple nodules who were treated withlobar TACE (P ¼ 0.172).Another interesting finding is the relationship

observed between tumor necrosis and the diameter ofthe nodule. We noted a significantly direct relationshipbetween necrosis and the tumor diameter: the greaterthe tumor diameter, the greater the percentage of ne-crosis. To correctly interpret these data, we shouldconsider that our patient population had small HCCnodules (all < 5 cm and almost all < 4 cm). Thus,21 of the 122 nodules of the complete series (mainlynodules 3-4 cm in size) showed the best response. It iswell known that larger HCCs are fed by larger arteries;this leads to better visualization of the nodule duringangiographic examination.31-33 The reverse side of thisissue is that smaller nodules (those that have just tran-sitioned from regenerative nodules with severe dyspla-sia to very early HCCs34) are more often hypovascular;in other words, in these nodules, arterial tumoral neo-angiogenesis is not fully developed, and they are stillnourished by a limited portal blood supply.32,34,35 Wepreviously reported36 that more than 15% of smallHCC nodules (1-3 cm) lack the typical contrast HCCpattern at imaging (hyperenhancement in the arterialphase followed by washout),1,2 and these theoreticallycorrespond to nodules in which vascular derangementhas not yet fully taken place; thus, the potential ofTACE is limited. In light of these considerations, ourfinding that the smaller nodules in the present serieswere less efficiently treated by arterial chemoemboliza-tion than the larger ones is not unexpected. Further-more, Alba and coworkers12 reported that tumors thatwere preoperatively detected by CT because of hyper-vascularity had more necrosis (mean ¼ 67.8%) andwere larger (2.58 cm) than those not detected preoper-atively (mean necrosis level ¼ 1.57%, diameter ¼1.68 cm). Riaz and coworkers28 found instead that thehighest rate of complete necrosis after TACE could beachieved in nodules in the intermediate size range of 3to 5 cm in comparison with smaller and larger nod-ules. Our population and consequent findings are dif-ferent from those patients undergoing TACE as aunique modality (usually with HCCs at an intermedi-ate stage) and their findings. These patients often havenodules > 5 cm and rarely have tumors < 2 cm; inthis case, an increase in the rate of necrosis in thelarger ones would be completely unexpected.Our multivariate logistic regression analysis showed

that the independent predictors of complete tumor ne-crosis were the selective/superselective TACE procedure

HEPATOLOGY, Vol. 53, No. 5, 2011 GOLFIERI ET AL. 1587

(P ¼ 0.049) and a single nodule treatment (P ¼0.008), whereas the nodule size played a minor role (P¼ 0.089).In the setting of a locoregional bridge treatment for

LT, the rate of complete necrosis according to a patho-logical analysis after percutaneous radiofrequency abla-tion was reported to be approximately 50%, but it was61% to 63% in nodules < 3 cm and 15% in HCCs> 3 cm (usually 3-5 cm).37 The rate of complete ne-crosis in the entire series was fully comparable to thatachieved after selective/superselective TACE in ourpresent population of transplant patients (53.8%).These data, together with our finding that HCCs 3 to5 cm in size had a higher rate of necrosis than smallertumors after TACE, support the strategy (commonlyused in many centers) of using a bridge treatmentinvolving percutaneous ablation for smaller nodules(<3 cm) and selective/superselective TACE for largernodules (>3 cm). Combined treatment might also bean option.As for the post-TACE CT assessment, dense Lipidol

uptake proved to be a poorly specific marker of a com-plete histological response. Dedicated trials are war-ranted to identify the best strategy for post-TACEevaluations.In conclusion, the present study shows that in small

HCCs (<5 cm), selective/superselective TACE is ableto achieve considerable rates of tumor necrosis compa-rable to the rates reported for HCCs treated with ra-diofrequency ablation in the same setting of transplantpatients; very small nodules (<2 cm), however, do notrespond as well as 3- to 4-cm nodules. If selective/su-perselective procedures cannot be technically per-formed, lobar procedures may then nonetheless beused, but in this situation, the expected rate of necrosishas been shown to be lower.

Acknowledgment: The authors thank their col-leagues in the Bologna Liver Transplant Group as wellas Emanuela Giampalma, Matteo Renzulli, and Cris-tina Mosconi (Radiology Unit, University of Bologna),who supported the management of the patients.

References1. Bruix J, Sherman M. Management of hepatocellular carcinoma.

HEPATOLOGY 2005;42:1208-1236.

2. Bruix J, Sherman M. Management of hepatocellular carcinoma: anupdate. http://www.aasld.org/practiceguidelines/Documents/Bookmar-ked%20practice%20Guidelines/Hccupdate2010.pdf. Accessed February2011.

3. Ravaioli M, Grazi GL, Piscaglia F, Trevisani F, Cescon M, Ercolani G,et al. Liver transplantation for hepatocellular carcinoma: results ofdown-staging in patients initially outside the Milan selection criteria.Am J Transplant 2008;8:2547-2557.

4. Llovet JM, Bruix J. Systematic review of randomized trials for unresect-able hepatocellular carcinoma: chemoembolization improves survival.HEPATOLOGY 2003;37:429-442.

5. Llovet JM, Real MI, Montana X, Planas R, Coll S, Aponte J, et al. Ar-terial embolisation or chemoembolisation versus symptomatic treatmentin patients with unresectable hepatocellular carcinoma: a randomisedcontrolled trial. Lancet 2002,18;359:1734-1739.

6. Marelli L, Stigliano R, Triantos C, Senzolo M, Cholongitas E, DaviesN, et al. Transarterial therapy for hepatocellular carcinoma: which tech-nique is more effective? A systematic review of cohort and randomizedstudies. Cardiovasc Intervent Radiol 2007;30:6-25.

7. Camma C, Schepis F, Orlando A, Albanese M, Shaied L, Trevisani F,et al. Transarterial chemoembolization for unresectable hepatocellularcarcinoma: meta-analysis of randomized controlled trials. Radiology2002;224:47-54.

8. Uchida H, Ohishi H, Matsuo N, Nishimine K, Ohue S, Nishimura Y,et al. Transcatheter hepatic segmental arterial embolization using Lipio-dol mixed with an anticancer drug and Gelfoam particles for hepatocel-lular carcinoma. Cardiovasc Intervent Radiol 1990;13:140-145.

9. Uchida H, Matsuo N, Sakaguchi H, Nagano N, Nishimine K, OhishiH. Segmental embolotherapy for hepatic cancer: keys to success. Cardi-ovasc Intervent Radiol 1993,16:67-71.

10. Matsui O, Kadoya M, Yoshikawa J, Gabata T, Arai K, Demachi H,et al. Small hepatocellular carcinoma: treatment with subsegmentaltrans-catheter arterial embolization. Radiology 1993;188:79-83.

11. Matsuo N, Uchida H, Nishimine K, Soda S, Oshima M, Nakano H,et al. Segmental transcatheter hepatic artery chemoembolization withiodized oil for hepatocellular carcinoma: antitumor effect and influenceon normal tissue. J Vasc Interv Radiol 1993;4:543-549.

12. Alba E, Valls C, Dominiguez J, Martinez J, Escdalante E, Llado L,et al. Transcatheter arterial chemoembolization in patients with hepato-cellular carcinoma on the waiting list for orthotopic liver transplanta-tion. AJR Am J Roentgenol 2008;190:1341-1348.

13. Piscaglia F, Camaggi V, Ravaioli M, Grazi GL, Zanello M, Leoni S,et al. A new priority policy for patients with hepatocellular carcinomaawaiting liver transplantation within the Model for End-Stage LiverDisease system. Liver Transpl 2007;13:857-866.

14. Freeman RB Jr, Wiesner RH, Harper A, McDiarmid SV, Lake J,Edwards E, et al. The new liver allocation system: moving towards evi-dence-based transplantation policy. Liver Transpl 2002;8:851-858.

15. United Network for Organ Sharing. http://www.unos.org. AccessedFebruary 2011.

16. Llovet JM, Bru C, Bruix J. Prognosis of hepatocellular carcinoma: theBCLC staging classification. Semin Liver Dis 1999;19:329-384.

17. Bruix J, Sherman M, Llovet JM, Beaugrand M, Lencioni R, BurroughsAK, et al. Clinical management of hepatocellular carcinoma. Conclu-sions of the Barcelona-2000 EASL conference. European Associationfor the Study of the Liver. J Hepatol 2001;35:421-430.

18. Piscaglia F, Gianstefani A, Ravaioli M, Golfieri R, Cappelli A, Giam-palma E, et al. Criteria for diagnosing benign portal vein thrombosis inthe assessment of patients with cirrhosis and hepatocellular carcinomafor liver transplantation. Liver Transpl 2010;16:658-667.

19. Pomfret EA, Washburn K, Wald C, Nalesnik MA, Douglas D, RussoM, et al. Report of a national conference on liver allocation in patientswith hepatocellular carcinoma in the United States. Liver Transpl 2010;16:262-278.

20. Raoul JL, Sangro B, Forner A, Mazzaferro V, Piscaglia F, Bolondi L,et al. Evolving strategies for the management of intermediate-stage he-patocellular carcinoma: available evidence and expert opinion on theuse of transarterial chemoembolization. Cancer Treat Rev; doi:10.1016/j.ctrv.2010.07.006.

21. Majno PE, Adam R, Bismuth H, Castaing D, Ariche A, Krissat J,et al. Influence of preoperative transarterial lipiodol chemoembolizationon resection and transplantation for hepatocellular carcinoma inpatients with cirrhosis. Ann Surg 1997;226:688-701.

22. Bargellini I, Vignali C, Cioni R, Petruzzi P, Cicorelli A, Campani D,et al. Hepatocellular carcinoma: CT for tumor response after

1588 GOLFIERI ET AL. HEPATOLOGY, May 2011

transarterial chemoembolization in patients exceeding Milan criteria—selection parameter for liver transplantation. Radiology 2010;255:289-300.

23. Oldhafer KJ, Chavan A, Fruhauf NR, Flemming P, Schlitt HJ, KubickaS, et al. Arterial chemoembolization before liver transplantation inpatients with hepatocellular carcinoma: marked tumor necrosis, but nosurvival benefit? J Hepatol 1998;29:953-959.

24. Decaens T, Roudot-Thoraval F, Bresson-Hadni S, Meyer C, Gugen-heim J, Durand F, et al. Impact of pretransplantation transarterial che-moembolization on survival and recurrence after liver transplantationfor hepatocellular carcinoma. Liver Transpl 2005;11:767-775.

25. Maddala YK, Stadheim L, Andrews JC, Burgart LJ, Rosen CB,Kremers WK, et al. Drop-out rates of patients with hepatocellular can-cer listed for liver transplantation: outcome with chemoembolization.Liver Transpl 2004;10:449-455.

26. Wong LL, Tanaka K, Lau L, Komura S. Pre-transplant treatment of he-patocellular carcinoma: assessment of tumor necrosis in explanted livers.Clin Transplant 2004;18:227-234.

27. Yao FY, Kinkhabwala M, LaBerge JM, Bass NM, Brown R Jr, KerlanR, et al. The impact of pre-operative loco-regional therapy on outcomeafter liver transplantation for hepatocellular carcinoma. Am J Trans-plant 2005;5:795-804.

28. Riaz A, Lewandowski RJ, Kulik L, Ryu RK, Mulcahy MF, Baker T,et al. Radiologic-pathologic correlation of hepatocellular carcinomatreated with chemoembolization. Cardiovasc Intervent Radiol 2010;33:1143-1152.

29. Belghiti J, Carr BI, Greig PD, Lencioni R, Poon RT. Treatment beforeliver transplantation for HCC. Ann Surg Oncol 2008;15:993-1000.

30. Dharancy S, Boitard J, Decaens T, Sergent G, Boleslawski E, DuvouxC, et al. Comparison of two techniques of transarterial chemoemboliza-tion before liver transplantation for hepatocellular carcinoma: a case-control study. Liver Transpl 2007;13:665-671.

31. Tajima T, Honda H, Taguchi K, Asayama Y, Kuroiwa T, Yoshimitsu K,et al. Sequential hemodynamic change in hepatocellular carcinoma anddysplastic nodules: CT angiography and pathologic correlation. AJRAm J Roentgenol 2002;178:885-897.

32. Kudo M. Multistep human hepatocarcinogenesis: correlation of imag-ing with pathology. J Gastroenterol 2009;44(Suppl 19):112-118.

33. Hayashi M, Matsui O, Ueda K, Kawamori Y, Kadoya M, Yoshikawa J,et al. Correlation between the blood supply and grade of malignancyof hepatocellular nodules associated with liver cirrhosis: evaluation byCT during intraarterial injection of contrast medium. AJR Am JRoentgenol 1999;172:969-976.

34. Matsui O. Imaging of multistep human hepatocarcinogenesis by CTduring intra-arterial contrast injection. Intervirology 2004;47:271-276.

35. Piscaglia F, Bolondi L. Recent advances in the diagnosis of hepatocellu-lar carcinoma. Hepatol Res 2007;37:S178-S192.

36. Leoni S, Piscaglia F, Golfieri R, Camaggi V, Vidili G, Pini P, et al. Theimpact of vascular and nonvascular findings on the noninvasive diagno-sis of small hepatocellular carcinoma based on the EASL and AASLDcriteria. Am J Gastroenterol 2010;105:599-609.

37. Pompili M, Mirante VG, Rondinara G, Fassati LR, Piscaglia F, AgnesS, et al. Percutaneous ablation procedures in patients with cirrhosiswith hepatocellular carcinoma submitted to liver transplantation: assess-ment of efficacy at explant analysis and of safety for tumor recurrence.Liver Transpl 2005;11:1117-1126.

HEPATOLOGY, Vol. 53, No. 5, 2011 GOLFIERI ET AL. 1589