Chemoembolization Follow-up ofHepatocellular Carcinoma with MR Imaging:Usefulness of Evaluating Enhancement Featureson One-month Posttherapy MR Imaging forPredicting Residual DiseaseBobby Kalb, MD, Abbas Chamsuddin, MD, Lama Nazzal, MD, Puneet Sharma, PhD,

and Diego R. Martin, MD, PhD

PURPOSE: To determine the sensitivity, specificity, and accuracy of contrast-enhanced magnetic resonance (MR) imagingperformed 1 month after localized chemotherapy as a measure of tumor response, before detectable changes in size.

MATERIALS AND METHODS: This trial was approved by the authors’ institutional review board and was compliantwith the Health Insurance Portability and Accountability Act (HIPAA). Inclusion criteria selected patients receivingchemoembolization for hepatocellular carcinoma (HCC) with MR imaging within 2 months before treatment, inaddition to MR imaging after treatment at 1 month and 6 months. Pathology was used as a surrogate for 6-monthfollow-up if the patient underwent interval transplantation. The final population consisted of 23 tumors (occurringwithin 21 patients). MR imaging studies were evaluated separately by two radiologists. Tumors were scored asshowing complete loss of enhancement or as showing some residual tissue enhancement. Changes in T1 and T2 signaland perilesional enhancement were tabulated and recorded. Lesion size was also measured on all MR imaging studiesby using a one-dimensional measure of the longest dimension. Increase in tumor size from 1–6 months of 20% orgreater was used as confirmation of residual disease. In 5 of 23 tumors, review of pathology served as the surrogatestandard. Sensitivity, specificity and accuracy were computed for each rater.

RESULTS: The sensitivity, specificity, and accuracy of 1-month follow-up MR imaging were 71.4–85.7%, 100%, and91.3–95.7%. There was a high degree of agreement between the two readers for both the 1-month (� � 0.88) and6-month (� � 1.0) MR imaging studies.

CONCLUSIONS: This investigation shows high accuracy for using tumor enhancement features on 1-month post-therapy MR imaging to predict residual disease after chemoembolization of HCC.

J Vasc Interv Radiol 2010; 21:1396–1404

Abbreviations: GRE � gradient-recalled echo, HCC � hepatocellular carcinoma, LC � low compression, RECIST � Response Evaluation Criteria in SolidTumors, T1W, T2W � T1-weighted, T2-weighed

HEPATOCELLULAR carcinoma (HCC)is a significant public heath issue andthe fifth leading cause of death world-

From the Department of Radiology (B.K., P.S.,D.R.M.), Emory University School of Medicine, 1365Clifton Road NE, Building A–AT622, Atlanta, GA30322; Center for Laser and Interventional Surgery(A.C.), Atlanta, Georgia; and Department of InternalMedicine (L.N.), St. Louis University Hospital, St.Louis, Missouri. Received December 11, 2009; final

revision received April 27, 2010; accepted May 11,

1396

wide (1). Liver transplantation is thetreatment of choice for early HCC thatmeets the Milan criteria (2); however,

2010. Address correspondence to D.R.M.; E-mail:dmartin@emory.edu

None of the authors have identified a conflict ofinterest.

© SIR, 2010

DOI: 10.1016/j.jvir.2010.05.015

transplantation has had less favorableresults in the setting of more advancedtumor (3). Alternative therapeuticmethods are needed for cases of ad-vanced tumors that are unsuitable fortransplantation or resection. Percuta-neous arterial embolization of tumorcombined with local chemotherapyhas emerged as a treatment option inthese cases. Transarterial chemoembo-lization has shown a clear benefit inthe setting of unresectable disease (4,5)

and has been used as neoadjuvant

Kalb et al • 1397Volume 21 Number 9

therapy before orthotopic liver trans-plantation and resection (6,7). More re-cently, drug-eluting microspheres thatare loaded with doxorubicin (low-compression [LC] beads) have beenused to provide a more targeted, sus-tained chemotherapeutic effect with adecreased systemic dose (8,9). LC beadtherapy differs from chemoemboliza-tion because blood supply to the tu-mor is not obliterated at the end of thedelivery procedure.

Assessing response to chemoembo-lization is important for guiding pa-tient management, and an incompletetumor response often prompts addi-tional, repeated percutaneous treat-ments. Treatment response is mostcommonly evaluated with either com-puted tomography (CT) or MR imag-ing, and the most commonly used in-dicator of treatment response is achange in tumor size. The ResponseEvaluation Criteria in Solid Tumors(RECIST) (10) are the most widelyused guidelines for evaluating tumorresponse to therapy, with change insize as the sole criterion for treatmentresponse. A change in tumor size aftertherapy may require several months tomanifest, however, and size changesmay be misleading in the early post-treatment setting because of changessecondary to tumor hemorrhage andnecrosis (11,12). For these reasons, moresensitive, specific, and rapid methodsfor documenting and measuring tumorresponse would be desirable to guideindividualized therapy.

HCCs are generally supplied by he-patic arterial blood flow and showcharacteristic intense enhancement onarterial phase contrast–enhanced CTor MR imaging (13). It has been sug-gested that tumor response may bedetermined by a change in enhance-ment on imaging, a feature adopted bythe European Association for theStudy of the Liver (14). Size evaluationstill remains the primary criterion,however, for determination of treat-ment response in most clinical trials(15–17). MR imaging has been used asthe imaging modality of choice forevaluating HCC at some centers (13)and provides superior sensitivity ofcontrast enhancement compared withCT (18). The aim of this study was todetermine the sensitivity, specificity,and accuracy of contrast-enhancedMR imaging performed 1 month after

localized chemotherapy as a measure

of tumor response before any detect-able changes in size.

MATERIALS AND METHODS

Patients

This trial was approved by our in-stitutional review board and was com-pliant with HIPAA. Prospective in-formed consent was obtained from allpatients undergoing MR imaging forposttherapy evaluation of HCC. Thisarticle was written in accordance withthe SIR Reporting Standards (19).

Inclusion criteria selected patientsthat had received chemoembolizationfor HCC between December 2005 andJanuary 2008 who had contrast-en-hanced MR imaging within 2 monthsbefore treatment, in addition to post-procedure MR imaging performed at 1and 6 months after treatment. The6-month posttherapy scan was used tomonitor tumor growth as evidence ofpersistent viable tumor. Alternatively,patients who had liver transplantationbefore the 6-month MR imaging studyhad explant liver pathologic analysis toevaluate for residual tumor. Patientswere excluded if any additional treat-ment was implemented during the6-month follow up period.

We identified 45 patients who hadpretreatment MR imaging and chemo-embolization of HCC. Pretreatment MRimaging enabled diagnosis of HCC us-ing criteria accepted by the United Net-work for Organ Sharing (20). Therewere 16 patients excluded secondary toinadequate long-term interval follow-up(� 6 months). An additional eight pa-tients were excluded because of multi-ple embolization treatments betweenMR imaging scans. The final study pop-ulation included 21 patients withchronic liver disease. Two of these pa-tients developed a second tumor in thecontralateral hepatic lobe and oppositeto the lobe initially treated. These twopatients had localized chemoemboliza-tion of the new tumors and were im-aged with the same protocol, bringingthe total to 21 patients with 23 tumors.

The final study cohort consisted of21 patients: 17 men and 4 women witha mean age of 61 years (range 51–78years). All patients had a clinical his-tory of chronic liver disease. The etiol-ogy of liver disease (some patientswith more than one primary cause)

included hepatitis C virus infection

(n � 15), hepatitis B virus infection(n � 2), alcohol-related causes (n � 6),and cryptogenic causes (n � 3).

Chemoembolization

Chemoembolization was performedwith doxorubicin-loaded LC bead em-bolization in all patients except three:two with conventional chemoemboliza-tion and one with Sirtex yttrium-90 (Y90)microsphere embolization. For LC beadtherapy, embolization was performedwith 300- to 500-� and 500- to 700-�doxorubicin-loaded LC beads. For thetwo cases of conventional chemoembo-lization, a mixture of ethiodized oil (10mL), doxorubicin (50 mg), cisplatin (100mg) and 300- to 500-�m polyvinyl alco-hol particles was injected for emboliza-tion. For radiopharmaceutical emboli-zation, 44 mCi of Y90-loaded Sirtexmicrospheres was injected.

Magnetic Resonance Imaging

All MR imaging studies were per-formed on 1.5T MR scanners (MagnetomAvanto, Siemens Medical Solutions, Ise-lin, New Jersey [SMS]; Gyroscan Intera,Philips Medical Systems, Einthoven,Netherlands [PMS]; or Twin-EXCITE, GEHealthcare, Milwaukee, Wisconsin [GE]).Patients scheduled on one vendor plat-form had subsequent follow-up scansscheduled on the same machine. A mul-tiple-channel torso phased array coil wasused for signal reception.

Our routine liver protocol consistsof a combination of single-shot T2-weighted (T2W) images and dynamic,contrast-enhanced sequences extend-ing from the lung bases through thekidneys. T2W images were acquiredwith a single-shot fast spin-echo se-quence in the coronal and axial planeswithout fat saturation and in the axialplane with fat saturation. A breath-hold (end inspiration), dual-echospoiled gradient-recalled echo (GRE)sequence was also obtained for evalu-ation of fat and iron content. Three-dimensional T1-weighted (T1W) GREimages were obtained in the precon-trast phase (axial and coronal) andsubsequently in the arterial, venous,and delayed phases with the followingparameters: 380 mm2 field of view, 288matrix (70% phase resolution), partialFourier imaging, TR/TE/flip angle �3.8/2.0/10 degrees, 96 slices at 3-mm

slice thickness, and bandwidth at 380

1398 • MRI Follow-up of HCC Post-chemoembolization September 2010 JVIR

Hz/pixel (alternatively, 6-mm slicethickness reconstructed at 3-mm in-tervals using twofold zero interpola-tion [General Electric]).

Each patient’s weight was recordedand gadobenate dimeglumine (Multi-Hance, Bracco Diagnostics, Inc., Prince-ton, New Jersey) was administered at adose of 0.05 mmol/kg and a rate of 2mL/sec, followed by a 30-mL salineflush at 2 mL/sec using a dual-chamberpower injector (Spectris, Medrad, War-rendale, Pennsylvania). Arterial phaseimages were acquired using a real-timebolus-tracking method, with breath-hold instructions and image acquisi-tion begun at the point the contrastagent bolus reaches the diaphragmand the celiac trunk. This method re-sults in reproducible, arterial phaseimages that are less sensitive to thepatient’s hemodynamic status. Acqui-sition time for three-dimensional GREimages was 17–18 sec and performedin a single breath hold. The time delaybetween bolus trigger point and acqui-sition initiation was 8 sec. Venousphase imaging was initiated at 70 secafter the bolus trigger point, providingadequate time for patient recoveryand a second breath hold. The delayedphase images were acquired at 180 secafter the bolus trigger point.

Magnetic Resonance ImagingAnalysis

All MR imaging data sets were trans-ferred to and evaluated on a separatecomputer with image review software(Efilm v3.0, Merge Healthcare). Two ra-diologists with 12 and 2 years of expe-rience in abdominal MR imaging inter-preted the 1-month posttherapy studieswhile blinded to the results of the fol-low-up 6-month posttherapy studies, al-though the pretherapy studies wereavailable for review. Review sessionswere performed independently and ondifferent days. Each tumor was scoredas either complete treatment response totherapy or as partial treatment responsewith residual tumor present.

The 6-month posttherapy scans wereinterpreted on a separate day, and eachreader had access to the 1-month fol-low-up MR imaging studies to evaluatefor treatment response. The 6-monthMR imaging studies were scored us-ing the same criteria used in the1-month posttherapy studies. The pa-

tients’ images were accessed, and the

results were recorded by a uniquemedical record number and not by pa-tient name to reduce the possibility ofreading bias.

Criteria for Therapeutic Response

The arterial and delayed phase im-ages were analyzed for tumor en-hancement. For simplicity and to testpracticality of the method, we usedvisual determination of changes in en-hancement, comparing the pretherapyand posttherapy arterial phase images.If the tumor was scored as showingcomplete loss of enhancement, thiswas interpreted as a complete re-sponse to therapy. If the tumor showedat least some partial or focal residualarterial enhancement with washout,this was interpreted as the presence ofresidual disease after therapy. Onlythe tissue within the tumor marginswas considered.

Additionally, changes in precon-trast T1 and T2 signals were noted byeach reviewer and tabulated for eachpatient as isointense, increased, or de-creased relative to adjacent liver pa-renchyma. Enhancement changes inthe liver adjacent to the tumor weresimilarly recorded, and any visuallydetectable change was evaluated foreach phase of contrast enhancement tonote in which phase the changes weremost conspicuous.

Size Measurements

Lesion size was measured for eachtumor during pretherapy and 1-monthMR imaging studies and for the non-explanted cases that had a 6-monthfollow-up study by using a one-di-mensional measure of the longest di-mension in keeping with RECIST cri-teria (10,12). The margin of the arterialenhancing tissue that showed washoutwas used for consistent marking of thetumor walls on pretherapy examina-tions. If residual arterial enhancing tis-sue was shown on the posttherapy ex-amination, this enhancing soft tissuewas measured in a single longest di-mension. The margins of an enhancingtumor on the arterial phase imagecould not be used in tumors that lostthis enhancement after therapy. Inthese cases, the treated tumor site wasmeasured and orientated to the samerelative positions as used for pre-

therapy tumor measurements.

Pathology

In 5 of the 23 tumors, a 6-month fol-low-up MR imaging study was unavail-able because the patient underwent ei-ther interval liver transplantation (n �4) or died and an autopsy was per-formed (n � 1) before this time. In thesecases, review of the pathology regard-ing the explant liver served as the sur-rogate standard of reference. Retro-spective correlation of the explantedtumor with the lesion shown on MRimaging was based on established he-patic segmental anatomic descriptions(Couinaud classification) to ensurethat the tumor evaluated on gross pa-thology corresponded with the lesionshown on MR imaging. The presenceof any viable tumor identified by thepathologist was scored as a case ofresidual disease, whereas the presenceof a necrotic lesion without viable tu-mor was scored as complete responseto therapy.

Statistical Analysis

The change in tumor size from the1-month to the 6-month follow-up MRimaging study (or pathologic evalua-tion) served as a standard of referencefor defining the presence or absence ofresidual disease against which the1-month MR imaging findings werecompared in all patients. Three catego-ries were defined: (1) An increase intumor size from 1–6 months of 20% orgreater was defined as confirmationof residual disease (in keeping withRECIST criteria and the known growthrate of HCC) (21,22). (2) Increase intumor size of 0–19% was defined asindeterminate. (3) Decrease in tumorsize from 1–6 months was defined asconfirmation of a complete response totherapy. Sensitivity, specificity, andaccuracy of 1-month follow-up MRimaging in predicting residual diseasewere computed along with 95% score-based confidence intervals for each rater.Interrater reliability of the 1-month and6-month readers was assessed withCohen’s � statistic. To investigate iftumor size changes between the pre-therapy scan and the 1-month fol-low-up correlated with the presence orabsence of residual disease, a rank

sum test was used.

um ent

Kalb et al • 1399Volume 21 Number 9

RESULTS

The results of image interpreta-tion for each reader are summarizedin Table 1. Sensitivity, specificity,and accuracy of 1-month follow-upMR imaging are presented in Table 2,with associated 95% confidence in-tervals.

Five treated tumors underwent in-terval transplantation before 6-monthfollow-up MR imaging could be ob-

Table 1Results of Posttherapy Magnetic Resona

TumorNo.

Findings at 1 mo* Fi

Reader 1 Reader 2 Rea

1 Neg (FN) Neg (FN) Neg2 Neg (TN) Neg (TN) Neg3 Neg (TN) Neg (TN) Neg4 Pos (TP) Pos (TP) Pos5 Neg (TN) Neg (TN) Neg6 Neg (TN) Neg (TN) Neg

7 Neg (TN) Neg (TN) Neg8 Neg (TN) Neg (TN) Neg9 Neg (TN) Neg (TN) Neg

10 Pos (TP) Pos (TP) Pos11 Neg (TN) Neg (TN) Neg12 Neg (TN) Neg (TN) Neg13 Neg (TN) Neg (TN) Neg14 Neg (TN) Neg (TN) Neg15 Pos (TP) Pos (TP) Pos16 Neg (TN) Neg (TN) Neg17 Neg (TN) Neg (TN) Neg

18 Pos (TP) Pos (TP) Pos19 Neg (TN) Neg (TN) Neg

20 Neg (TN) Neg (TN) Neg21 Pos (TP) Neg (FN) Pos22 Pos (TP) Pos (TP) Pos23 Neg (TN) Neg (TN) Neg

* Neg � complete loss of arterial phase efalse positive, TN � true negative, TP �† Neg � decreased tumor size, Pos � inc‡ Neg � no viable tumor within treated t

Table 2Statistics for Magnetic Resonance Imagi

Sensitivity (%);

Reader 1 85.7 (6/7); [4Reader 2 71.4 (5/7); [3

Note—Numbers in parentheses are raw dCI � confidence interval.

tained. Three of the five explanted tu-

mors showed residual disease, and theremaining two tumors showed a com-pletely necrotic treatment site on path-ologic evaluation. Anatomic correspon-dence between the explant liver tumorsite and MR imaging localization wasconsidered highly confident in allcases. The 1-month follow-up MR im-aging scan identified correspondingpresence or absence of residual en-hancing tumor tissue in all but onecase: One patient in whom the tumor

Imaging Interpretations of Readers 1 an

ings at 6 mo* Size Changefrom 1–6mo (%)

Growth6 mo1 Reader 2

) Neg (FN)) Neg (TN) �14.3 Neg) Neg (TN)

) Pos (TP)) Neg (TN) �16.7 Neg) Neg (TN)

) Neg (TN) �8.7 Neg) Neg (TN) �33.3 Neg) Neg (TN) �38.8 Neg

) Pos (TP) �26.7 Pos) Neg (TN) �11.8 Neg) Neg (TN) �20.0 Neg) Neg (TN) �20.0 Neg) Neg (TN) �9.3 Neg

) Pos (TP) �66.7 Pos) Neg (TN) �10.4 Neg) Neg (TN) �16.7 Neg

) Pos (TP) �44.4 Pos) Neg (TN) �25.0 Neg

) Neg (TN) �23.1 Neg) Pos (TP) �24.2 Pos) Pos (TP)

) Neg (TN) �27.8 Neg

ncement, Pos � persistent arterial phase ee positive.sed tumor size � 20%.or bed; Pos � viable tumor within treatm

Detection of Residual Tumor on 1-month

% CI] Specificity (%); [95%

7] 100 (16/16); [81–102] 100 (16/16); [81–10

.

was interpreted by both readers as a

case of complete loss of enhancementon the 1-month follow-up examina-tion eventually underwent liver trans-plantation 2.9 months after the initialfollow-up examination. Pathologicevaluation in this case showed twosmall foci of HCC (each measuring � 5mm) at the periphery of a large area ofinfarcted tissue. A retrospective sec-ond interpretation by both reviewersof this treated area on the 1-monthfollow-up examination confirmed no

at 1 and 6 Months

ExplantPathology‡ Treatment

Pos LC beadsLC beads

Neg LC beadsPos LC beads

LC beadsNeg 90Y Sirtex

microspheresLC beadsLC beadsLC beadsLC beadsLC beadsLC beadsLC beadsLC beadsLC beadsLC beadsTransarterial

chemoembolizationLC beadsTransarterial

chemoembolizationLC beadsLC beads

Pos LC beadsLC beads

ancement, FN � false negative, FP �

bed.

llow-up Examination

] Accuracy (%); [95% CI]

95.7 (22/23); [79–99]91.3 (21/23); [73–98]

nce d 2

ndat

†der

(FN(TN(TN

(TP(TN(TN

(TN(TN(TN

(TP(TN(TN(TN(TN

(TP(TN(TN

(TP(TN

(TN(TP(TP(TN

nha nhtrurea

ng Fo

[95 CI

9–9 0]6–9 0]

ata

foci of arterial enhancement to suggest

1400 • MRI Follow-up of HCC Post-chemoembolization September 2010 JVIR

imaging findings of residual enhanc-ing disease.

For the remaining 18 tumors inwhich no explant data existed, the sizechanges from 1-month to 6-month fol-low-up MR imaging served as thestandard of reference. Using sizechange at 6 months as the standard ofreference, Reader 1 identified persis-tent posttherapy tumor enhancementthat correlated with subsequent tumorgrowth on follow-up MR imaging orcomplete loss of enhancement thatcorrelated with decreased tumor sizein 16 of 16 patients and 18 of 18 tu-mors. Reader 2 reported the same en-hancement findings with the same cor-relations in 15 of 16 patients and 17 of18 tumors.

Overall, using a combination ofpathologic analysis and size changesat 6 months as the standards of refer-ence, Reader 1 correctly identified thepresence or absence of residual dis-ease in 20 of 21 patients and 22 of 23

Table 3Signal Intensity Changes* and Perilesio

TumorNo.

T1 (Precontrast)

Pretherapy 1-mo Postt

1 � �2 � �3 � �4 � �5 � �6 iso �7 iso �8 � �9 iso �

10 � �11 � iso12 � iso13 � iso14 iso �15 � �16 � �17 � �18 iso �19 � �20 � �21 � �22 � �23 � iso

* Visual assessment of signal intensity rel� � hypointense to liver.† Presence (�) or absence (�) of nonmasstumor.‡ Extensive portal venous clot was presenexamination.

tumors, whereas Reader 2 correctly

identified 19 of 21 patients and 21 of 23tumors (Table 1).

The 1-month follow-up MR imag-ing scan showed excellent accuracy(Table 2) for predicting residual dis-ease. There was a high degree ofagreement between the two readersfor the 1-month and 6-month MR im-aging studies. Results for image inter-pretation at 6 months were identical(ie, complete agreement) for the tworeaders. At 1 month, there was highinterobserver reliability with � � 0.88(95% confidence interval 0.65–1). Allpredictions for both readers were com-pletely specific, and accuracies rangedfrom 79–96%.

Size Measurements

Changes in tumor size from pre-therapy MR imaging to 1-month MRimaging were not significantly differ-ent (P � .20) in patients who exhibitedinterval growth at 6 months (mean

Enhancement before and after Therapy†

T2

apy Pretherapy 1-mo Postthera

� �� �iso iso� �� iso� �� �� �� �� �iso isoiso isoiso iso� iso� iso� �iso iso� �� �iso iso� �iso iso� �

e to adjacent liver parenchyma. � � hype

e perilesional enhancement in the hepatic

this case, contributing to marked abnorm

change of �0.54 cm), compared with

patients who showed decreased sizeof the treated tumor at 6 months(mean change of �0.43 cm). Tumorsize ranged from 1.2–11.0 cm, and themean diameter for the study popula-tion was 3.4 cm. Although not a pri-mary aim of this study, enhancementfeatures of the 6-month follow-up MRimaging scan showed 100% correla-tion with size changes at 6 months forboth readers. Tumors that showed im-aging features of persistent enhancingtissue on the 6-month follow-up exam-ination (n � 4) showed a correspond-ing increase in size of the residual en-hancing tumor compared with the1-month posttherapy examination (av-erage size increase �40.5%, range�24.2–�66.7%). Tumors that showedno findings for residual disease on the6-month posttherapy examination(n � 14) showed a corresponding de-crease in size of the treated tumor (av-erage size decrease �19.7%, range

Perilesional enhancement

Pretherapy 1-mo Posttherapy

� �� �� �� �� �� �� ��‡ �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �

tense to liver, iso � isointense to liver,

enchyma surrounding the treated

perilesional perfusion on pretherapy

nal

her py

ativ rin

lik par

t in al

�8.7–�38.8%). No tumors showed

gro na

Kalb et al • 1401Volume 21 Number 9

growth rates in the indeterminate (0–20% growth) category (Table 1).

Signal Features

Qualitative precontrast T1 signal

Figure 1. Residual tumor after chemoemPrecontrast MR imaging: precontrast (a), ar(arrows) appearing on a background of chrthree-dimensional GRE (d), arterial phasetreatment bed on precontrast imaging (arrothe treatment site (arrowheads in e), in kee(arrow in e). MR imaging 6 months postth(i) images show an increase in the size of aimaging (i), confirming residual viable and

intensity, T2 signal intensity, and per-

ilesional enhancement were docu-mented for each case (Table 3). Over-all, 61% (14 of 23) of treated tumorsshowed an increase in T1 signal rela-tive to the pretherapy examination,and 26% (6 of 23) showed stable T1

lization with precontrast and postcontrasal phase (b), and delayed phase (c) images

c liver disease, in keeping with HCC. MR imand delayed phase (f) images show the d

n d) but with residual nodules of arterial eg with residual disease. Reactive, perilesiopy: precontrast T1W three-dimensional Gial enhancing tissue (h) within the treatmewing tumor corresponding to the tissue rem

signal. Of tumors, 70% (16 of 23)

showed stable T2 signal, and the re-maining 30% (7 of 23) manifested de-creased T2 signal after therapy. Peril-esional enhancement was present innearly all cases (21 of 23, 91%) of post-therapy tumors; only 1 case (4%)

1W three-dimensional GRE MR imaging.w arterial enhancing tumor with washouting 1 month posttherapy: precontrast T1Welopment of intrinsic T1 signal within thencing tissue along the posterior margin ofenhancement has developed after therapy(g), arterial phase (h), and delayed phaseite, which also shows washout on delayednts that have persistent vascularization.

bo t Tteri shooni ag(e), evw i nhapin nalera RE

rter nt s

showed perilesional enhancement on

1402 • MRI Follow-up of HCC Post-chemoembolization September 2010 JVIR

Figure 2. Complete response to therapy. Pretherapy MR imaging: Precontrast T1W three-dimensional GRE (a), arterial phase (b),delayed phase (c), and single-shot T2W (d) images show HCC in segment 6 (arrows). MR imaging 1 month posttherapy: precontrast

T1W three-dimensional GRE (e), arterial phase (f), delayed phase (g), and single-shot T2W (h) images show complete loss of enhancing

Kalb et al • 1403Volume 21 Number 9

pretherapy examination. This patientalso had extensive portal thrombosis,which is a probable etiologic factor forthe surrounding altered perfusion onthe pretherapy examination in thiscase.

DISCUSSION

This study investigated the useful-ness of early (1-month) posttherapycontrast-enhanced MR imaging for thedetection of residual disease after che-moembolization for HCC. Our resultsshow good sensitivity for the ability ofthe 1-month follow-up MR imagingstudy to predict the presence of resid-ual disease using persistent arterialphase contrast enhancement as amarker of viable tumor (Fig 1). In ad-dition, there was 100% specificity forboth readers in predicting completeresponse to therapy (Fig 2). Therewas overall excellent accuracy of the1-month posttherapy MR imagingscan in evaluating response to ther-apy. No significant change was foundin size after therapy at 1 month be-tween the tumors that increased insize and the tumors that becamesmaller at 6 months (P � .20). Theclinical significance of these findings isthat a 1-month posttherapy contrast-enhanced MR imaging study can pro-vide earlier reporting of treatmenteffectiveness, compared with the useof change in tumor size alone.

Other changes were identified 1month after therapy related to tumorsignal on T1W and T2W unenhancedimages. These changes (Figs 1 and 2),although not specific for tumor re-sponse, were useful as markers forprior treatment and include elevatedT1W signal intensity (61%). The mostreliable marker of prior therapy wasthe development of increased peril-esional enhancement (enhancementoutside the borders of the treated tu-mor), which was consistently noted onposttherapy examinations (91%) andnot on pretherapy examinations andwas presumably due to reactive he-patic parenchymal changes inducedby chemoembolization effects.

tumor tissue (arrowheads in f). Increased TT2 signal has decreased after therapy (h). M(j), delayed phase (k), and single-shot T2W

in the interim. Reactive, perilesional enhancem

Percutaneous therapies, such astransarterial chemoembolization, havebeen shown to improve survival in pa-tients with unresectable HCC (5). Dataregarding the benefit of chemoemboli-zation before liver transplantation areless clear (23), but it may provide abenefit when liver transplantation isperformed during a specified intervalafter therapy (23) and with repeatedtreatments. In addition, patients withtumor characteristics exceeding theMilan criteria may subsequently bedownstaged to allow listing for livertransplantation (6). Size measurementchanges are currently the most widelyused method for determining thera-peutic response; however, our resultsshow that enhancement features arelikely more reliable than size changesalone in the early posttreatment set-ting. Earlier determination of thera-peutic response may be helpful toguide patient management decisionsafter chemoembolization, especiallyregarding the need for repeat percuta-neous treatment.

We used a size threshold increaseof 20% or greater from 1–6 months asthe standard for determining residualdisease. This is in keeping with sizethresholds that are routinely used forevaluating disease response, specifi-cally with standardized RECIST crite-ria. In addition, the biology of HCC issuch that the tumor tends to grow rap-idly, and recurrent HCC specificallyhas been shown to have a tumor vol-ume doubling time that is significantlyshorter than primary HCC, rangingfrom 70–75 days in two different stud-ies (21,22).

An important potential limitationin interpreting the findings related todecreased posttherapy enhancementin our study is that the loss of arterialphase tumor enhancement could re-flect the results of embolization ther-apy and arterial occlusion. If the lossof enhancement was purely a measureof embolization by the locally deliv-ered intraarterial beads, however, wewould expect that the residual tumorenhancement with chemoemboliza-tion would be more complete than

gnal has developed within the treatment simaging 6 months posttherapy: precontrastimages confirm no residual enhancing tissu

ent changes surround the margins of the tre

with LC beads, where tumors retainperfusion even immediately after ther-apeutic delivery (ie, LC bead therapydepends on chemoelution locally de-livered, not on complete tumor embo-lization). Although one concern maybe that our study patients had morethan one type of therapy, most of ourpatients had LC bead therapy, and pa-tients who showed tumor response on6-month follow-up or on histologyalso show showed complete loss ofarterial tumor enhancement. Thesefindings would be more in keepingwith tumor response leading to invo-lution of tumor vessels and decreasedarterial perfusion within 1 month oftherapy.

We used observational methodol-ogy with qualitative measures ofchanges in enhancement and signal in-tensities within the tumor. It may bethat quantitative measures could pro-vide superior evaluation of tumor en-hancement changes. Use of quantita-tive measures would require morehighly developed software for spatialregistration of the tumor and sur-rounding tissues than is routinelyavailable currently, to allow registra-tion of precontrast and postcontrastimage sets. Higher temporal resolu-tion, allowing for acquiring multipleimages during the perfusion phase ofenhancement, may also provide moredetailed analysis of tumor vascularchanges. This would potentially addcomplexity to the scan, however, anddiminish the diagnostic value of theoverall arterial phase imaging that isroutinely acquired. Although thesepotential improvements should beconsidered, we intentionally focusedon routinely available scanning method-ology and used simple observationalanalysis. The fact that our results show ahigh degree of accuracy for determiningtumor response serves as a measure ofthe utility of routinely available meth-odology.

In one patient, the explant livercomparative pathology acquired 2.9months after therapy showed 5-mmnodular foci of residual tumor tissuein an otherwise necrotic tumor. This

n precontrast images (arrow in e), whereasthree-dimensional GRE (i), arterial phase

ithin treatment bed, which has contracted

1 si te oR i T1W(l) e w

ated tumor (arrowhead in j).

1404 • MRI Follow-up of HCC Post-chemoembolization September 2010 JVIR

apparent false-negative MR imagingstudy at 1 month after therapy mayhave been due to an inability to detecta small volume of residual tumor be-cause of spatial resolution limits,growth of the tumor nodules duringthe interval between MR imaging andsurgery, or a combination of these fac-tors.

Another limitation is that we haveevaluated tumor response by variousmeasures, but we do not have long-term evaluation of patient outcomes.This limitation is partly a result of abroadly inclusive study populationthat included patients meeting trans-plant criteria (intent to cure) and pa-tients not meeting transplant criteria(palliative intervention). We cannotyet state that patients with tumor re-sponse, by any measure, have differ-ent, presumably better, outcomes thanpatients who do not have tumor re-sponse, either with or without subse-quent liver transplantation. An in-creased patient sample size combinedwith a longer follow-up time would berequired to make such a statement. Wealso do not have follow-up on any pos-sible patients who had additional inter-ventions 1–6 months after the first treat-ment and cannot state whether therewas any possible bias in our studygroup as a result. By using a broadlyinclusive study population, however,we believed that this would provide amore rapid patient accrual rate duringthe term of study and minimize poten-tial preselection bias by including pa-tients with an array of tumor biology.

In conclusion, with the increasingrole of chemoembolization in the ther-apy of unresectable HCC and as anadjunct to preliver transplantationtherapy, early determination of thera-peutic response is important to guidecontinuing treatment in individual pa-tients. Analysis of tumor enhancementfeatures with MR imaging within 1month of therapy, using routinelyavailable methods, is a potentially use-ful and accurate predictor of therapeu-tic response.

References1. Llovet JM, Burroughs A, Bruix J. He-

patocellular carcinoma. Lancet 2003;

362:1907–1917.

2. Mazzaferro V, Regalia E, Doci R, et al.Liver transplantation for the treatmentof small hepatocellular carcinomas inpatients with cirrhosis. N Engl J Med1996; 334:693–699.

3. Yokoyama I, Todo S, Iwatsuki S, StarzlTE. Liver transplantation in the treat-ment of primary liver cancer. Hepato-gastroenterology 1990; 37:188–193.

4. Lesurtel M, Mullhaupt B, Pestalozzi BC,Pfammatter T, Clavien PA. Trans-arterial chemoembolization as a bridgeto liver transplantation for hepatocel-lular carcinoma: an evidence-basedanalysis. Am J Transplant 2006; 6:2644–2650.

5. Lo CM, Ngan H, Tso WK, et al.Randomized controlled trial of transar-terial lipiodol chemoembolization forunresectable hepatocellular carcinoma.Hepatology (Baltimore, Md) 2002; 35:1164–1171.

6. Schwartz M, Roayaie S, Uva P.Treatment of HCC in patients awaitingliver transplantation. Am J Transplant2007; 7:1875–1881.

7. Otto G, Heise M, Moench C, et al.Transarterial chemoembolization be-fore liver transplantation in 60 patientswith hepatocellular carcinoma. Trans-plant Proc 2007; 39:537–539.

8. Kalva SP, Thabet A, Wicky S. Recentadvances in transarterial therapy ofprimary and secondary liver malignan-cies. RadioGraphics 2008; 28:101–117.

9. Varela M, Real MI, Burrel M, et al.Chemoembolization of hepatocellularcarcinoma with drug eluting beads: ef-ficacy and doxorubicin pharmacoki-netics. J Hepatol 2007; 46:474–481.

10. Therasse P, Arbuck SG, Eisenhauer EA,et al. New guidelines to evaluate theresponse to treatment in solid tumors.European Organization for Researchand Treatment of Cancer, NationalCancer Institute of the United States,National Cancer Institute of Canada.J Natl Cancer Inst 2000; 92:205–216.

11. Forner A, Ayuso C, Varela M, et al.Evaluation of tumor response after lo-coregional therapies in hepatocellularcarcinoma: are response evaluation cri-teria in solid tumors reliable? Cancer2009; 115:616–623.

12. Suzuki C, Jacobsson H, Hatschek T, etal. Radiologic measurements of tu-mor response to treatment: practicalapproaches and limitations. Radio-Graphics 2008; 28:329–344.

13. Lauenstein TC, Salman K, Morreira R,et al. Gadolinium-enhanced MRI fortumor surveillance before liver trans-plantation: center-based experience.AJR Am J Roentgenol 2007; 189:663–

670.

14. Bruix J, Sherman M, Llovet JM, et al.Clinical management of hepatocellularcarcinoma: conclusions of the Barce-lona-2000 EASL conference. EuropeanAssociation for the Study of the Liver.J Hepatol 2001; 35:421–430.

15. Herber SC, Otto G, Schneider J, et al.Transarterial chemoembolization inpatients not eligible for liver transplan-tation: single-center results. AJR Am JRoentgenol 2008; 190:1035–1042.

16. Thomas MB, Morris JS, Chadha R, et al.Phase II trial of the combination of be-vacizumab and erlotinib in patientswho have advanced hepatocellular car-cinoma. J Clin Oncol 2009; 27:843–850.

17. Ko HK, Ko GY, Yoon HK, Sung KB.Tumor response to transcatheter arte-rial chemoembolization in recurrenthepatocellular carcinoma after livingdonor liver transplantation. Korean JRadiol 2007; 8:320–327.

18. Hunt SJ, Yu W, Weintraub J, PrinceMR, Kothary N. Radiologic monitor-ing of hepatocellular carcinoma tumorviability after transhepatic arterial che-moembolization: estimating the accu-racy of contrast-enhanced cross-sec-tional imaging with histopathologiccorrelation. J Vasc Interv Radiol 2009;20:30–38.

19. Brown DB, Gould JE, Gervais DA, et al.Transcatheter therapy for hepatic ma-lignancy: standardization of terminol-ogy and reporting criteria. J Vasc In-terv Radiol 2007; 18:1469–1478.

20. Freeman RB Jr, Wiesner RH, Harper A,et al. The new liver allocation system:moving toward evidence-based trans-plantation policy. Liver Transpl 2002;8:851–858.

21. Tezuka M, Hayashi K, Kubota K, et al.Growth rate of locally recurrent hepa-tocellular carcinoma after transcatheterarterial chemoembolization: compar-ing the growth rate of locally recurrenttumor with that of primary hepatocel-lular carcinoma. Dig Dis Sci 2007; 52:783–788.

22. Park Y, Choi D, Lim HK, et al.Growth rate of new hepatocellular car-cinoma after percutaneous radiofre-quency ablation: evaluation with mul-tiphase CT. AJR Am J Roentgenol 2008;191:215–220.

23. Aloia TA, Adam R, Samuel D, AzoulayD, Castaing D. A decision analysismodel identifies the interval of efficacyfor transarterial chemoembolization(TACE) in cirrhotic patients with hep-atocellular carcinoma awaiting livertransplantation. J Gastrointest Surg

2007; 11:1328–1332.