www.elsevier.com/locate/jhep

Journal of Hepatology 49 (2008) 498–501

Editorial

Molecular tools and hepatocellular carcinoma: Adding helpor confusion in liver transplantation? q

Eric Vibert, Didier Samuel*

Inserm, Unite 785, Villejuif, F-94800, France

Univ Paris-Sud, UMR-S 785, Villejuif, F-94800, France

AP-HP Hopital Paul Brousse, Centre Hepato-Biliaire, Villejuif, France

See Article, pages 581–588

Liver transplantation (LT) is in theory, the best treat-ment for hepatocellular carcinoma (HCC) in cirrhoticpatients since it is a treatment that simultaneously dealswith the malignancy and its causes [1–3]. However, theprognosis is directly related to the risk of HCC recur-rence post-transplantation. The aim of preoperativeassessment in LT for HCC is therefore to detect existingextra-hepatic metastases and predict extra-hepaticmicro-metastases that are responsible for recurrence,graft loss and often, death of the patient.

Following the retrospective study by Bismuth et al.[4], Mazzaferro et al. prospectively showed that carefulselection of patients with a single lesion of <5 cm orthree lesions each <3 cm, known as Milan criteria, hasresulted in a satisfactory balance between survival(75%) at four years and recurrence rates (8%) [5]. Thevalidity of these criteria to predict recurrence-free sur-vival after transplantation has been largely confirmedby several studies, which have demonstrated an overall10-year survival ranging from 60% to 62% [6,7]. Hence,survival of carefully selected patients is not differentfrom survival of patients transplanted for non-malig-nant indications [8]. Survival analysis by ITT clearlysuggested that resection or ablation in patients withcompensated cirrhosis are the only alternatives to liver

0168-8278/$34.00 � 2008 European Association for the Study of the Liver.

doi:10.1016/j.jhep.2008.07.004

Associate Editor: M. Colomboq The authors declare that they do not have anything to disclose

regarding funding from industries or conflict of interest with respect tothis manuscript.

* Corresponding author. Tel.: +33 1 45 59 34 03; fax: +33 1 45 59 3857.

E-mail address: didier.samuel@pbr.aphp.fr (D. Samuel).

transplantation on the wait list. Moreover there is grow-ing evidence that by limiting OLT only to patients fall-ing within the Milan criteria might represent an unjustpenalty for selected patients beyond the Milan criteriawho could still have a favourable outcome with OLT.

In the United States, around 70% of cirrhotic patientsdiagnosed with a HCC fall beyond the Milan criteria [3].Survival analysis by ITT clearly suggested that regularultrasound screening of cirrhotic patients has permitteddetection of small unique nodule in Child-Pugh A cir-rhotic patients in whom alternatives to transplantationare possible. Second, liver transplantation will providea major survival gain in some patients with a HCCwho fall beyond the Milan criteria. Therefore, severalexpansions of the Milan criteria have been proposed,with the University of California, San Francisco(UCSF) criteria described by Yao et al. being the mostinvestigated [9,10]. The UCSF criteria include a solitarytumor of 6.5 cm or less, no more than three lesions withthe largest being 4.5 cm or less, and a total tumor diam-eter 8 cm or less, without gross vascular invasion.Unfortunately, the 5-year survival of patients beyondMilan and within UCSF criteria ranges from 46% to93% [9–12]. The resulting discrepancies between studieswhich used UCSF criteria are related to the time ofapplication of the criteria (preoperative/radiologicaland postoperative/pathological). Interestingly, a Frenchmulticentric study of 479 patients demonstrated a higherrate of understaging (48%) if UCSF criteria is used tolist patients for LT as compared to the understaging rateof 34% when the Milan criteria is used [11]. Indeed, the5-year survival of UCSF excluding Milan patients was

Published by Elsevier B.V. All rights reserved.

E. Vibert, D. Samuel / Journal of Hepatology 49 (2008) 498–501 499

45% compared to 60% in Milan patients [11]. Heteroge-neous survival results in studies using UCSF criteriaemphasized that macroscopic parameters like size andnumber would probably lose their efficacy as surrogatecriteria to accurately predict the risk of tumor recur-rence in comparison to tumoral volume exceeding theMilan criteria.

Several studies have identified vascular invasion asthe strongest independent predictor factor of recurrence,a pathological parameter also associated with otherpoor prognostic factors like a poor differentiation ofthe tumor, the presence of satellite nodules and a highlevel of a-fetoprotein [6,13,14]. Unfortunately, exceptfor the latter, all these parameters are not easily avail-able in the preoperative setting and particularly atlisting.

In 2003, based on the knowledge that HCC displaymolecular alterations Marsh et al., classified HCCaccording to the level of gene damage [15]. The goal ofthis paper was to determine whether a panel of tumorsuppressor gene markers could be useful to predictrecurrence after LT for HCC. By microdissection,non-neoplastic and neoplastic tissue samples for eachpatient transplanted with a minimum of 5 years fol-low-up were evaluated for informative status of 18 geneslocated on microsatellite. Absence of microsatellites orloss of heterozygoty (LOH) corresponded to the lossof specific tumor suppressor genes such as APC,CDKN2A, p53, and p34. Among 18 microsatellitestested, 9 were significantly correlated with the post-transplant recurrence-free survival but none achievedperfect discrimination alone. As a test, two types of sim-plified panel of LOH that were the most accurate to pre-dict recurrence were defined. Thanks to this method andwith the help of an artificial neural network, an accurateresponse was obtained in 91 of the 103 patients tested(88%). The prediction of recurrence was accurate in89% of cases.

Very recently, the same team has performed a similaranalysis on a larger cohort of patients transplanted forHCC with more than 5 years of follow-up (183 patientsfrom Mount Sinai and University of Pittsburgh MedicalCenter) [16]. In this study, a fractional allelic imbalance(FAI) rate was calculated for each patient. This indexwas defined as the number of mutated markers dividedby the total number of informative markers. Thesemarkers corresponded to the 9 microsatellites retrievedas correlated with the post-transplant recurrence freesurvival in their initial work [15]. Of note, all these 9markers were not always informative, i.e present, in eachpatient. The rate of mutation, i.e. FAI, was then depen-dent on the number of informative markers. Hence, theFAI represents a rough estimate of cumulative muta-tional damage in tumor compared to non-neoplastic tis-sue. FAI in the 9-gene panel ranged from 0% to 89%(mean 31%). Among the patients tested, FAI and vascu-

lar invasion were the most important independent fac-tors of tumor-free survival. Patients with FAI > 40%compared to those with FAI 6 20% have an increasedrisk of recurrence by a factor of 19.5. Comparatively,patients with macrovascular invasion compared withthose without vascular invasion have an increased riskof recurrence of 5.9. This difference emphasized thepower of the FAI even comparatively with the more pre-dictive pathological factor, vascular invasion [6,13,14].Interestingly, a correlation with the Milan, UCSF crite-ria and FAI has been described in this paper. Forpatients outside the Milan criteria, the risk of recurrenceincreased 4.5-fold compared to patients within, while aFAI > 40% increased this risk 14-fold compared toFAI 6 20%. For patients beyond UCSF criteria, the riskof recurrence increased 6.0-fold compared to patientswithin, while a FAI > 40% increased this risk 17.4-foldcompared to FAI 6 20%. In summary, this secondpaper has demonstrated the much more powerful corre-lation with recurrence of FAI compared with patholog-ical data and macroscopic usual classification.

Hence, the paper by Schwartz et al. which appears inthis issue of the Journal of Hepatology [17] focused onthe potential applicability of the FAI score in patientsbeyond the Milan criteria with a favourable outcomeof LT. With the same methods of allelic imbalance anal-ysis of HCC in transplanted patients, they have com-pared tumoral and non-tumoral liver tissue for 18genes markers to determine the FAI score. Statistically,a FAI score of 0.27 was determined as the more reliablecut-off value to predict recurrence or not. Univariateand multivariate analysis of prognostic factors of recur-rence were performed with 70 patients transplanted forHCC with at least 5 years of follow-up. The FAI scoreand macroscopic vascular invasion were the 2 indepen-dent factors associated with recurrence. Interestingly,FAI predicted recurrence in 86% compared to 76% usingMilan criteria. Unfortunately, the statistical significanceif any of this difference was not reported in the paper.Besides, it was probably due to the low rate of recur-rence in Milan criteria (3/35) patients that this paperfocused on patients beyond Milan criteria. In the sub-group of patients beyond Milan criteria, the same pre-dictive factors of recurrence were confirmed but, inthis group, a FAI P 0.27 was predictive of recurrencewith a sensitivity of 83%, a specificity of 91% and overallaccuracy of 89%. Thus, this paper emphasized the utilityof FAI analysis only in patients beyond the Milan crite-ria. Besides, among the 35 patients within the Milan cri-teria, 4 of 6 of those who were predicted to haverecurrence according to FAI did not.

Even if it was useful in patients beyond Milan crite-ria, the major drawback of this technique was the neces-sity of full pathological examination. To date, all papersthat explored evaluation of FAI have used surgical spec-imens to predict recurrence. As underlined by Marsh

500 E. Vibert, D. Samuel / Journal of Hepatology 49 (2008) 498–501

et al. in 2003 [15], the two principal limitations in extrap-olating this work to the preoperative area were: (1) Alltumors are not recognized preoperatively. (2) It is notpractical or feasible to biopsy all detected tumors multi-ple times in the preoperative setting. Indeed, Pawliket al. have analyzed the concordance between histologi-cal data using preoperative needle core biopsy (14–18gauges) of HCC and final pathological analysis [18]. Inthis study that included 120 patients, preoperativebiopsy misclassified tumor grade in 55% of cases. Thepoor diagnostic accuracy of preoperative biopsy wasrelated to the very heterogeneous pathological andmolecular alteration of HCC, even in small tumors. Inci-dence of grading heterogeneity was reported in 18–46%[19–21] of HCC smaller than 2 cm and 36–43% in 2–3 cm nodules [19,21]. This heterogeneity was due tosequential development of HCC often displaying thenodule-in-nodule pattern where a subpopulation witha more malignant potential develops in a pre-existingnodule. Interestingly, as grading heterogeneity, molecu-lar alterations, notably mutation of p53 and b-catenin,were also retrieved inside small HCCs [21].

As reported in the paper by Schwartz et al. [17], thevalue of FAI was dependent on the number of mutatedgenes to the total number of informative genes. It wasthe variability of informative gene numbers that hasimposed the calculation of an index. Indeed, a recentpaper has focused on the genetic diversity of humanHCC [22]. In this work, a transcriptome analysis withgenotype and phenotype correlation has been performedin a surgical series of 120 HCCs and 3 hepatocellularadenomas. The transcriptome analysis identified 6 sub-groups (G1–G6) of HCCs associated with clinical andgenetic characteristics. HCC of G1 and G2 group wereassociated with HBV infection. G3 HCC mainlyincluded p53 mutation but without HBV infection. G4HCC was a heterogeneous group of HCCs.G5 and G6were highly related to b-catenin activation. Interestingly,HCC groups G1 to G3 were associated with a high rateof chromosomal instability in contrast to G5 and G6where no chromosome deletion, then no LOH was iden-tified. Hence, claims that FAI, largely correlated withLOH, in these subgroups of HCC is questionable andemphasized the fact that the FAI ratio is only a roughmeasure of cumulative mutational damage. It shouldbe emphasized that the correlation of FAI and viral hep-atitis was not assessed in the paper by Schwartz andcolleagues.

Despite these limitations, FAI represents a promisingnew tool to predict recurrence after LT, even if it is onlyof benefit postoperatively in patients falling beyond theMilan criteria. In the post-transplant setting, FAI couldprovide useful information to adapt immunosuppressivetherapy and adjuvant chemotherapy thus attenuatingthe risk of recurrence [23]. In the pre-transplant setting,the application of molecular markers needs further eval-

uation before being used as selection criteria for livertransplantation.

Acknowledgement

The authors thank Dr. Emir Hoti for reading themanuscript.

References

[1] Adam R, McMaster P, O’Grady JG, Castaing D, Klempnauer JL,Jamieson N, et al. Evolution of liver transplantation in Europe:report of the European Liver Transplant Registry. Liver Transpl2003;9:1231–1243.

[2] Yoo HY, Patt CH, Geschwind JF, Thuluvath PJ. The outcome ofliver transplantation in patients with hepatocellular carcinoma inthe United States between 1988 and 2001: 5-year survival hasimproved significantly with time. J Clin Oncol 2003;21:4329–4335.

[3] El-Serag HB, Siegel AB, Davila JA, Shaib YH, Cayton-WoodyM, McBride R, et al. Treatment and outcomes of treating ofhepatocellular carcinoma among medicare recipients in the UnitedStates: a population-based study. J Hepatol 2006;44:158–166.

[4] Bismuth H, Chiche L, Adam R, Castaing D, Diamond T, DennisonA. Liver resection versus transplantation for hepatocellular carci-noma in cirrhotic patients. Ann Surg 1993;218:145–151.

[5] Mazzaferro V, Chun YS, Poon RT, Schwartz ME, Yao FY,Marsh JW, et al. Liver transplantation for hepatocellular carci-noma. Ann Surg Oncol 2008;15:1001–1007.

[6] Jonas S, Bechstein WO, Steinmuller T, Herrmann M, Radke C,Berg T, et al. Vascular invasion and histopathologic gradingdetermine outcome after liver transplantation for hepatocellularcarcinoma in cirrhosis. Hepatology 2001;33:1080–1086.

[7] Zavaglia C, De Carlis L, Alberti AB, Minola E, Belli LS, SlimAO, et al. Predictors of long-term survival after liver transplan-tation for hepatocellular carcinoma. Am J Gastroenterol2005;100:2708–2716.

[8] Mazzaferro V, Regalia E, Doci R, Andreola S, Pulvirenti A,Bozzetti F, et al. Liver transplantation for the treatment of smallhepatocellular carcinomas in patients with cirrhosis. N Engl JMed 1996;334:693–699.

[9] Yao FY, Ferrell L, Bass NM, Watson JJ, Bacchetti P, Venook A,et al. Liver transplantation for hepatocellular carcinoma: expan-sion of the tumor size limits does not adversely impact survival.Hepatology 2001;33:1394–1403.

[10] Yao FY, Xiao L, Bass NM, Kerlan R, Ascher NL, Roberts JP.Liver transplantation for hepatocellular carcinoma: validation ofthe UCSF-expanded criteria based on preoperative imaging. Am JTransplant 2007;7:2587–2596.

[11] Decaens T, Roudot-Thoraval F, Hadni-Bresson S, Meyer C,Gugenheim J, Durand F, et al. Impact of UCSF criteria accordingto pre- and post-OLT tumor features: analysis of 479 patientslisted for HCC with a short waiting time. Liver Transpl2006;12:1761–1769.

[12] Duffy JP, Vardanian A, Benjamin E, Watson M, Farmer DG,Ghobrial RM, et al. Liver transplantation criteria for hepatocel-lular carcinoma should be expanded: a 22-year experience with467 patients at UCLA. Ann Surg 2007;246:502–509.

[13] Pawlik TM, Delman KA, Vauthey JN, Nagorney DM, Ng IO,Ikai I, et al. Tumor size predicts vascular invasion and histologicgrade: implications for selection of surgical treatment for hepa-tocellular carcinoma. Liver Transpl 2005;11:1086–1092.

[14] Parfitt JR, Marotta P, Alghamdi M, Wall W, Khakhar A, SuskinNG, et al. Recurrent hepatocellular carcinoma after transplanta-tion: use of a pathological score on explanted livers to predictrecurrence. Liver Transpl 2007;13:543–551.

E. Vibert, D. Samuel / Journal of Hepatology 49 (2008) 498–501 501

[15] Marsh JW, Finkelstein SD, Demetris AJ, Swalsky PA, SasatomiE, Bandos A, et al. Genotyping of hepatocellular carcinoma inliver transplant recipients adds predictive power for determiningrecurrence-free survival. Liver Transpl 2003;9:664–671.

[16] Dvorchik I, Schwartz M, Fiel MI, Finkelstein SD, Marsh JW.Fractional allelic imbalance could allow for the development of anequitable transplant selection policy for patients with hepatocel-lular carcinoma. Liver Transpl 2008;14:443–450.

[17] Schwartz M, Dvorchik I, Roayaie S, Fiel MI, Finkelstein S,Marsh JW, et al. Liver transplantation for hepatocellularcarcinoma: Extension of indications based on molecular markers.J Hepatol 2008;49:581–588.

[18] Pawlik TM, Gleisner AL, Anders RA, Assumpcao L, Maley W,Choti MA. Preoperative assessment of hepatocellular carcinomatumor grade using needle biopsy. Implications for transplanteligibility. Ann Surg 2007;245:435–442.

[19] Kenmochi K, Sugihara S, Kojiro M. Relationship of histologicgrade of hepatocellular carcinoma (HCC) to tumor size, and

demonstration of tumor cells of multiple different grades in singlesmall HCC. Liver 1987;7:18–26.

[20] Nakashima O, Sugihara S, Kage M, Kojiro M. Pathomorphologiccharacteristics of small hepatocellular carcinoma: a special refer-ence to small hepatocellular carcinoma with indistinct margins.Hepatology 1995;22:101–105.

[21] An FQ, Matsuda M, Fujii H, Tang RF, Amemiya H, Dai YM,et al. Tumor heterogeneity in small hepatocellular carcinoma:analysis of tumor cell proliferation, expression and mutation ofp53 and beta-catenin. Int J Cancer 2001;93:468–474.

[22] Boyault S, Rickman DS, de Reynies A, Balabaud C, RebouissouS, Jeannot E, et al. Transcriptome classification of HCC is relatedto gene alterations and to new therapeutic targets. Hepatology2007;45:42–52.

[23] Zimmerman MA, Trotter JF, Wachs M, Bak T, Campsen J,Skibba A, et al. Sirolimus-based immunosuppression followingliver transplantation for hepatocellular carcinoma. Liver Transpl2008;14:633–638.