Supportla Reche

Accepte

ReprintColorecHopital(APHP)du G�enpanis@b

0039-60

� 2011

doi:10.1

496 S

Selective portal vein ligation andembolization induce different tumoralresponses in the rat liverL�eon Maggiori, MD,a Fr�ed�eric Bretagnol, MD, PhD,a Annie Sibert, MD,b Val�erie Paradis, MD, PhD,c

Val�erie Vilgrain, MD, PhD,b and Yves Panis, MD, PhD,a Clichy, France

Background. Portal vein ligation (PVL) and portal vein embolization (PVE) are used to enhance livervolume before hepatectomy for colorectal liver metastasis (LM). Impact of such techniques on tumorgrowth is not well known. This experimental study aimed to assess impact of PVE and PVL on LMgrowth in a murine model of colorectal LM.Methods. Single macroscopic tumor was induced by injection of 0.5 3 106 DHD/K12 cells under theliver capsule of BDIX rats at day 0. Multiple microscopic tumors were obtained by intra-portal injectionof 13 106 cells at day 7. At day 8, rats were divided in 3 groups: PVE group (selective 70% PVE), PVLgroup (selective 70% PVL); control group (sham laparotomy). Rats were sacrificed at day 37 (11 inPVE, 12 in PVL, and 10 rats in control groups). Liver volume and LM volumes were assessed.Results. Nonoccluded liver volume was larger in the PVE and PVL groups vs control group (P < .0001and P < .0001, respectively) but showed no difference in PVE vs PVL groups (P = .08). LM volume inthe occluded liver was smaller in the PVE vs control groups (P = .006) and larger in the PVL vs controlgroups (P = .001). LM volume in the nonoccluded liver was larger in the PVE and PVL groups vscontrol group (P = .010 and P = .010, respectively) but showed no difference in PVE vs PVL groups(P = .878).Conclusion. Both PVL and PVE modify tumor growth, especially in nonoccluded lobe. These results couldbe of clinical importance in humans where both techniques are widely used. (Surgery 2011;149:496-503.)

From the Department of Colorectal Surgery,a Department of Radiology,b and Department of Pathology,c

Beaujon Hospital, Assistance Publique-Hopitaux de Paris (AP-HP), Clichy, France

DESPITE MAJOR RECENT ADVANCE IN MEDICAL TREATMENTS,1

surgical resection provides the only chance forcure in patients with colorectal liver metastases(LM). The 5-year survival rate drops from approx-imately 30% after curative resection2,3 to less than5% if surgical treatment is not possible.4 There-fore, resection remains one of the cornerstonesof LM management.

Potentially curative surgical treatment can onlybe proposed to a third of the patients with LM,mainly because of inadequate future remnant liver

ed in part by research grants from the Association pourrche contre le Cancer.

d for publication October 19, 2010.

requests: Yves Panis, MD, PhD, Service de Chirurgietale, Pole des Maladies de l’Appareil Digestif (PMAD),Beaujon, Assistance Publique des Hopitaux de Paris, Universit�e Paris VII (Denis Diderot), 100 boulevard�eral Leclerc, 92118 Clichy cedex, France. E-mail: yves.jn.aphp.fr.

60/$ - see front matter

Mosby, Inc. All rights reserved.

016/j.surg.2010.10.012

URGERY

volume after hepatectomy. Portal vein occlusion(PVO) have been successfully proposed beforemajor hepatectomy to induce hypertrophy of thefuture remnant liver to minimize the risks ofpostoperative, life-threatening severe liver failure.

To date, 2 techniques of PVO have been de-scribed: portal vein embolization (PVE)5-7 and por-tal vein ligation (PVL).8 Comparative efficacies onliver hypertrophy of these 2 techniques have beenthe subject of debates in the literature.9-12 A recentstudy,9 however, suggests that PVE and PVL areequally efficient in terms of contralateral liverhypertrophy.

Other than the positive effects of PVO on futureliver function after major hepatectomy, few studieshave highlighted the potential of PVO to promotetumor growth, especially in the hypertrophied partof the liver that is not resected.13-17 Elias et al16

demonstrated that, after PVE, tumor growth ofLM was more rapid than the hypertrophy of thesurrounding liver parenchyma. Kokudo et al17

noted that PVE increased the tumor volume by ap-proximately 20% and was probably associated withan enhanced recurrence of disease.

Fig 1. Experimental murine model of bilobar colorectalliver metastases with portal vein occlusion of the left liver(the median and left lateral lobes: 70% of the total livervolume).

SurgeryVolume 149, Number 4

Maggiori et al 497

Thus, PVO might increase the risk of tumorprogression in the nonoccluded part of the liverfor patients with bilobar LM. This risk is even moreworrisome because even modern imaging proce-dures have a sensitivity of less than 50% for thediagnosis of LM #1 cm.18 This lack of sensitivitymeans that initial tests might not detect LM inthe nonoccluded liver that could be affected byPVO.

To our knowledge, no study has focused on thecomparison of tumor growth induced by PVE vsPVL. This comparison is difficult to assess in theclinical setting because of the extensive use ofneoadjuvant chemotherapy, which has been dem-onstrated to improve the progression-free survivalafter resection of LM.1 Therefore, the aim of thisexperimental study was to assess and comparethe effects of PVE and PVL on the growth rate ofLM in a relevant murine model of colorectal LM.

MATERIALS AND METHODS

Animals. The experiments were performed onmale BDIX rats (Charles River, Charles River, StGermain Sur L Arbresle, France), weighing 180 to220 g.The animalsweremaintained in a temperature-controlledenvironmentona12-hour light/dark cycleand given access to standard chow pellets and tapwater ad libitum. All procedures were performed incompliance with institutional animal care guidelines.

Cell line. DHD/K12 is a chemically induced ratcolon carcinoma cell line from BDIX rats.19 TheDHD/K12 cells were cultured in Dulbecco’s modi-fied Eagle’s medium and 1:1 Ham’s F-10 medium,containing 10% fetal calf serum. All media werecomplemented with 4 mM of L-glutamine, 100 UI/mL of penicillin, and 100 mg/mL of streptomycin.All materials used to culture the cell line were pur-chased from GIBCO, Invitrogen, Carlsbad, CA.

Culturing was realized in 150-cm2 culture flasks(Techno Plastic Products [TPP], Trasadingen,Switzerland) at 378Cand5%CO2.Whencells reached80% confluence, the medium was discarded, andthe cells were washed 3 times with Dulbecco’sPhosphate-Buffered Saline (DPBS;GIBCO, Invitro-gen) before incubation with 4 mL of trypsin-EDTAsolution (trypsin 0.05% and EDTA 0.2g/L;GIBCO, Invitrogen) at 378C for 5 minutes. Cellswere then harvested, centrifuged, counted, andre-suspended in DPBS for introduction to animals.

Operative procedures. The rats were anesthe-tized with intraperitoneal injections of pentobar-bital (0.001mL/g). During anesthesia, the ratswere allowed to breathe spontaneously. All opera-tive procedures were performed through a midlinelaparotomy that was closed with a single running

suture. After the operative procedure, the rats weremaintained 30 minutes under a heating lamp.

Colorectal liver metastasis model. Single mac-roscopic tumors were induced via direct injectionsof 0.5 3 106 DHD/K12 cells, which were sus-pended in 0.1 mL of DPBS under the liver capsuleof the median lobe using a 30-gauge needle. Mul-tiple microscopic tumors were obtained by injec-tion of 1 3 106 DHD/K12 cells into the portalvein using a 30-gauge needle (Fig 1).

Portal vein ligation. Selective portal vein liga-tion was performed on the left liver (the medianand left lateral lobes) corresponding to 70% of thetotal liver volume (Fig 1). After anatomy assess-ment and careful dissection of the hepatic artery,the corresponding portal veins were ligated witha 3-0 monofilament, nonabsorbable suture.

PVE. For PVE, the portal vein was puncturedwith a 20-gauge catheter. The catheter was con-nected with a 3-way stopcock and a 1-mL syringefilled with 5% glucose solution to prevent airbubble intrusion. The catheter was then placedjust after the origin of the right portal branches toallow selective embolization of the left liver (themedian and the left lateral lobes) correspondingto 70% of the total liver volume (Fig 1). Selectiveembolization was realized using a 0.15 mL solutionof cyanoacrylate and metacrylaxysulfolan (Glubran2; GEM, Viareggio, Italy) diluted 1:1 in 480 mg/mL iodized oil (Lipiodol; Guerbet, Roissy, France).After successful embolization, the catheter wasflushed with 0.2 mL of a 5% glucose solution.

Experimental protocols. All experimental pro-tocols were performed once but not repeated forverification.

SurgeryApril 2011

498 Maggiori et al

Preliminary protocol for PVO: At day 0, a total of 25rats were studied: the PVE group (n = 20), whichreceived selective PVE of the left liver (the medianand left lateral lobes), and the PVL group (n = 5),which received selective PVL of the left liver (themedian and left lateral lobes). A portography us-ing the contrast medium iohexol (Omnipaque;GE Healthcare, Piscataway, NJ) was performed be-fore PVO to control liver anatomy; the test was re-peated after PVO to demonstrate portal occlusionof appropriate liver segments. All rats were killedand autopsied at day 29. This preliminary protocolwas designed to obtain standardized techniques ofPVE and PVL.

Main protocol: At day 0, a total of 40 rats receiveddirect injections of 0.5 3 106 DHD/K12 cells un-der the liver capsule of the median lobe. At day7, all rats received injections of 1 3 106 DHD/K12 cells into the portal vein. At day 8, rats were di-vided into 3 groups: (1) the PVE group (n = 16),which received selective PVE of the left liver (themedian and left lateral lobes); (2) the PVL group(n = 12), which received selective PVL of the leftliver (the median and left lateral lobes); and(3) the control group (n = 12), which received asham midline laparotomy. All rats were killed andautopsied at day 37.

Pathologic evaluation. Immediately after death,the rats’ livers were explanted and split into right(including the superior and inferior right lobesand the anterior and posterior caudate lobes)and left (including the left lateral and medianlobes) sections. The right and left liver volumeswere evaluated by immersion into water with agraduated tube. The totality of each liver sectionwas fixed in 10% formaldehyde, and then embed-ded, sectioned, and stained with hematoxylin, eo-sin, and safran. Next, each section was analyzedwith computer-assisted optical microscopy. Eachmetastasis was identified and measured on its max-imal cut surface using Histolab software (Microvi-sion Instruments, Evry, France).

Tumor volume was assessed assimilating each

metastasis to a regular sphere using the following

formula: Volume ¼ 43Surface

ffiffiffiffiffiffiffiffiffiffiSurface

p

q. If more than

1 metastasis was found, the tumor volume was rep-resented by the sum of all the metastases volumes.All histological analyses were performed in ablinded fashion with respect to the experimentalgroups.

Statistical analysis. Continuous data are ex-pressed as mean ± standard deviation and com-pared with the Kruskal-Wallis test followed by apost hoc Mann-Whitney U test with the Bonferroni

correction to compensate for multiple compari-sons. Proportions are expressed as number of cases(percentage of cases) and compared with the Pear-son chi-square test or the Fisher exact test, as ap-propriate. Tests were always 2-sided and, withoutuse of the Bonferroni correction, the level of statis-tical significance was set at P < .05. Statistical anal-ysis was performed using the Statistical Package forthe Social Sciences (SPSS) software (version 16.0;SPSS Inc., Chicago, IL).

RESULTS

Preliminary protocol. A total of 5 rats wereexcluded from the PVE group because of death atanesthetic induction (n = 1) and post–PVE deathdue to an unknown cause after a mean delay of 3days (n = 4). For analysis, the PVE group and thecontrol group included 15 and 5 rats, respectively.

PVE group: At day 0, the mean weight of the ratswas 193 ± 7 g (range, 184–206 g). Portographyperformed after PVE showed that embolizationwas evident only in the left liver (the median andleft lateral lobes) in all cases. At day 29, a total of15 rats were killed and autopsied. Pathologicexamination showed that the mean right livervolume was 59% ± 2% (range, 56–63%) of thetotal liver volume.

PVL group: At day 0, the mean weight of the ratswas 195 ± 7 g (range, 183–207 g). Portographyperformed after PVL demonstrated that ligationwas evident only in the portal branches of the leftliver (the median and left lateral lobes) in all cases(Fig 2). At day 29, a total of 5 rats were killed andautopsied. Pathologic examination showed thatthe mean right liver volume was 60% ± 3% (range,56–64%) of the total liver volume.

Main protocol. Among all 40 rats, 7 (17.5%)animals were excluded from the study. A total of 4rats were excluded from the PVE group due todeath at anesthetic induction (n = 1) and post–PVE death after a mean delay of 2 days (n = 3).A total of 2 rats were excluded from the PVL groupdue to death at anesthetic induction, and 1 rat wasexcluded from the control group due to death atthe anesthetic induction. For analysis, the PVE,the PVL, and the control groups included 12, 10,and 11 rats, respectively.

Control group: At day 0, the mean weight of therats was 192 ± 7 g (range, 182–205 g). At day 37, atotal of 11 rats were killed and autopsied. No ratpresented peritoneal carcinomatosis or lung me-tastasis. Pathologic examination showed that themean right liver volume was 32% ± 4% (range,15–38%) of the total liver volume. Left LM werefound in all cases, and the mean left liver tumor

Fig 3. Right liver volume after portal vein embolization(PVE) or portal vein ligation (PVL) of the left liver inBDIX rats.

Fig 2. Selective portal vein ligation (PVL) of the left liverin BDIX rats. Portography results (A) before PVL and(B) after PVL.

Table. Liver volume and tumor volumes after portal vein embolization (PVE) or portal vein ligation (PVL)of the left liver in BDIX rats

PVE* N = 12 PVL* N = 10

Control* N = 11 P valuey P valueyRat weight, g 192 ± 7 196 ± 6 NS 194 ± 8 NSRight liver volume, % of total liver

volume32 ± 4 62 ± 7 <.0001 59 ± 5 <.0001

Left liver tumor volume, mm3 248 ± 9 128 ± 20 .006 565 ± 233 .001Right liver tumor volume, mm3 5 ± 8 17 ± 9 .010 26 ± 22 .010

*Values are given as mean ± standard deviation.yThe P values represent the PVE or the PVL group vs the control group.

NS, Not significant.

SurgeryVolume 149, Number 4

Maggiori et al 499

volume was 248 ± 9 mm3 (range, 123–451 mm3).Of the 11 rats in the control group, right LMwere found in 8 (73%), and the mean right livertumor volume was 5 ± 8 mm3 (range, 0.3–23 mm3).

PVE group: At day 0, the mean weight of the ratswas 196 ± 6 g (range, 187–205 g). At day 37, a totalof 12 rats were killed and autopsied. No ratpresented peritoneal carcinomatosis or lung me-tastasis. Pathologic examination showed that themean right liver volume was 62% ± 7% (range,44–70%) of the total liver volume. Left LM werefound in all cases, and the left liver mean tumorvolume was 128 ± 20 mm3 (range, 8–375 mm3). Ofthe 12 rats in the PVE group, right LM were foundin 8 (67%), and the right liver mean tumor volumewas 17 ± 9 mm3 (range, 2–34 mm3).

PVL group: At day 0, the mean weight of the ratswas 194 ± 8 g (range, 182–207 g). At day 37, a totalof 10 rats were killed and autopsied. No ratpresented peritoneal carcinomatosis or lung me-tastasis. Pathologic examination showed that themean right liver volume was 59% ± 5% (range,50–69%) of the total liver volume. Left LM were

Fig 5. Right liver tumor volume after portal vein embo-lization (PVE) or portal vein ligation (PVL) of the leftliver (the median and left lateral lobes) in BDIX rats.

Fig 4. Left liver tumor volume after portal vein emboli-zation (PVE) or portal vein ligation (PVL) of the leftliver in BDIX rats.

SurgeryApril 2011

500 Maggiori et al

found in all cases, and the left liver mean tumorvolume was 565 ± 233 mm3 (range, 169–936 mm3).Of the 10 rats in the PVL group, right LM werefound in 8 (80%), and the right liver mean tumorvolume was 26 ± 22 mm3 (range, 4–55 mm3).

Comparative analysis. As shown in the Table,the mean weight of the rats at day 0 showed no dif-ference among the 3 groups (P = .305). The meanvolume of the right liver (ie, the nonoccludedliver) was significantly different among the 3groups (P = .013, Fig 3). This volume was greaterin the PVE and PVL groups compared to the con-trol group (P < .0001 and P < .0001, respectively),but showed no difference in the PVE group vsthe PVL group (P = .08).

The mean volume of the left liver (ie, theoccluded liver) was significantly different amongthe 3 groups (P < .0001, Fig 4). This volume wasless in the PVE group compared to the controlgroup (P = .006) and more in the PVL group com-pared to the control group (P = .001).

The mean tumor volume of the right liver (ie,the nonoccluded liver) was significantly differentamong the 3 groups (P = .013, Fig 5). This volumewas greater in the PVE and PVL groups comparedto the control group (P = .010 and P = .010, respec-tively), but showed no difference in the PVE groupvs the PVL group (P = .878).

DISCUSSION

This experimental study was conducted to assessand compare the impact of PVE and PVL ontumor growth of colorectal LM. First, the dataconfirmed that both techniques are equally effec-tive in terms of contralateral liver hypertrophy.Second, the results suggest that both PVE and PVLhad a marked impact on tumor growth. The most

important result of this study is that both tech-niques increased tumor growth in the nonoc-cluded liver.

The experimental in vivo model used in thisstudy includes the association of the DHD/K12cell line and male BDIX rats. DHD/K12 is anestablished transplantable colon carcinoma cellline originating from a chemically induced coloncancer in BDIX rats.19 Other types of associationhave been reported with rats20 or with other spe-cies, such as mice or miniature pigs.21 The associ-ation of the DHD/K12 cell line and BDIX rats,however, presents several advantages for its utiliza-tion as an LM experimental model. First, BDIXrats have a normal immune system, which allowsconsistent study of natural history and treatmentsof LM. Second, a recent experimental study dem-onstrated that DHD/K12 introduced into BDIXrats provides the biologically most accurate invivo model of early colonic liver metastasisdevelopment.20

Moreover, we chose to induce LM by injectingtumor cells directly into the portal vein. Comparedto the injection of tumor cells into the spleen,which completely bypasses the natural history ofLM, the intraportal injection model mimics thenatural metastatic process, ie, the invasion oftumor cells through the portal vein. Moreover,the risk of splenectomy associated with the effectsof splenic injection could add an immunologicbias in the tumor growth study. Finally, comparedto other LM models, such as the intracecal model,the intraportal model produces up to 100% of LM6 weeks after the injection of cells.

The tumoral model used in this study is com-plex, including a single macroscopic LM in theoccluded liver, obtained by direct injection of

SurgeryVolume 149, Number 4

Maggiori et al 501

DHD/K12 cells under the liver capsule, and mul-tiple microscopic LM in both occluded and non-occluded livers, obtained by intraportal injectionof DHD/K12 cells. This model was used to mimic,as closely as possible, the clinical situation of apatient with a right liver metastasis and a leftremnant liver with undetectable micrometastases.In the majority of cases, PVO is used to treatpatients before a major right hepatectomy for anestablished right LM. However, left LM (in thenonoccluded lobe) might be present and remainundetected by the standard imaging procedure.18

Clinical observation of patients presenting withearly recurrences within the remnant liver aftermajor hepatectomy suggest the possible presenceof undetectable microscopic ‘‘dormant’’ LM thathave been positively stimulated by liver regener-ation either after hepatectomy or by the immu-nosupression induced by the laparotomy itself.22

In these situations, it is of clinical importanceto assess whether PVO could modify or stop tu-mor growth in the future remnant liver.

In humans, occlusion of greater than 50% ofthe portal flow is needed to induce a satisfactorycontralateral liver hypertrophy.23 In the presentstudy, an occlusion of 70% of the portal flow wasrealized, according to the same portal territory asthe largely described 70% hepatectomy in rat.24

The absence of statistical difference concerningthe level of hypertrophy between PVE and PVLobserved in this study is consistent with other clin-ical9,12 and experimental21,25 reports.

Several materials have been described for therealization of PVE: biologic glue,26 alcohol,27

microspheres,25,28 and cyanoacrylate.29 Furreret al25 recently compared the impact of PVE andPVL on liver regeneration in a rat model usingmicrospheres. They concluded that PVL was supe-rior to PVE in inducing a regenerative response ofthe remnant liver. Furrer et al,25 however, empha-sized that the impairment of liver regeneration ob-served after PVE might be a consequence ofmacrophage trapping in the occluded liver dueto a massive foreign body reaction induced bythe microspheres.

An immunologic response such as the one thatmight have been caused by microspheres in thestudy by Furrer et al25 could be different with otherembolizing materials. Cyanoacrylate, which wasused in this study, has demonstrated its efficacy toinduce liver regeneration after PVE.6 It remainsone the most popular materials and is commonlyused in clinical practice in our high-volume center.

Very few studies have focused on the impact ofPVO on tumor growth. Elias et al16 were the first to

show an acceleration of tumor growth after PVE bydemonstrating that LM growth wasmore rapid thanthe surrounding parenchyma in the nonoccludedliver. Several more recent clinical studies13-15,17 con-firmed the existence of an acceleration of tumorgrowth in the nonoccluded liver after PVE. Kokudoet al17 reported a significant increase of the meantumor volume in the nonoccluded part of the liverfrom 223 mL (prior to PVE) to 270 mL (3 weeksafter PVE).

Pamecha et al13 compared tumor growth rateper day in PVE vs control groups. They reporteda significant increase of this rate after PVE(0.36 mL per day) compared to a control group(0.05 mL per day). An experimental studyperformed by Heinrich et al30 compared the im-pact of 70% hepatectomy and PVL in a mouseLM model; they demonstrated an accelerationof tumor growth that was greater after PVLthan after hepatectomy. Our study, however, tothe best of our knowledge, is the first to directlycompare the impact of PVL and PVE on tumorgrowth.

Little information about the mechanisms oftumor growth is available; apparently, however,the tumor growth after PVL would be notcontrolled by the same mechanisms that controlhypertrophy of the nonligated or nonembolizedportions of the liver. Several factors could con-trol the tumor growth after portal occlusion;hemodynamic changes in blood supply of LM,for example, could explain the growth of LM.31

Because LM depend solely on arterial bloodflow, increased hepatic arterial flow induced bypartial vein obstruction could provide a poten-tial benefit for tumor growth. Moreover, PVLmay induce changes in secretion of cytokinesor growth factors such as hepatocyte growthfactor.32

We observed that PVE and PVL induced differ-ent tumoral response in the occluded liver: PVEwas associated with a reduction of tumor growth,whereas PVL was associated with an acceleratedtumor growth. This result could be explained byportal modifications induced by both techniquesthat are not strictly identical.11,33 PVL induces theformation of early portoportal collateral vessels,which are not observed after PVE.11 An increasein blood support through these collateral vesselsmight enhance tumor growth. In clinical practice,such accelerated tumor growth in the occludedliver after PVL could have potentially 2 negative ef-fects: the risk of contraindication of liver resection(ie, the occluded liver) and the risk of a worse on-cologic outcome (ie, the tumor spreading outside

SurgeryApril 2011

502 Maggiori et al

the occluded liver during tumor growth andbefore surgery).

In this study, we reported a similar increasedtumor growth in the nonoccluded liver after PVLand PVE compared to a control group. Suchacceleration could have a major clinical impacton the patient, jeopardizing further curative sur-gical resection. Ribero et al34 demonstrated thatapproximately 10% of patients did not undergocurative surgery after PVE for colorectal LM be-cause of disease progression. More recently, Pame-cha et al35 evaluated this rate and found it to be33% of the cases in their study. A metaanalysisalso demonstrated that only 85% of the patientsunderwent a laparotomy for attempted hepatec-tomy after PVE.36

Furthermore, an increasing tumor growth ratein the nonoccluded liver after PVO might have animpact on local recurrence and long-term survival.In the study by Pamecha et al,35 the 5-year overallsurvival rate was significantly lower if a PVE was re-alized before surgical resection of colorectal LM.A second study by Kokudo et al17 also reportedthat 2- and 4-year disease-free survival rates werelower after PVE than without it.

PVE and PVL have largely contributed to theextent of surgical indication for LM. However,results of the present study suggest caution regard-ing their use. The demonstration of an increasingtumor growth rate in the nonoccluded liver mightsupport the strategy of a 2-step surgical proce-dure.37 In patients with bilobar LM, the first stepcould be a left tumoral clearance associated witheither right PVE or PVL, followed by a right hepa-tectomy 6 weeks later. Finally, a recent study byBeal et al38 highlighted the interest of a peri–PVE chemotherapy to minimize disease progres-sion. This study demonstrated a reduction of thetumoral volume in the nonoccluded liver thatwas, nevertheless, associated with a reduction ofthe liver hypertrophy induced.38

In conclusion, our study suggests that both PVEand PVL have a significant impact on tumorgrowth of LM in the rat liver. In the occludedpart of the liver, PVL induced an increase in tumorgrowth, whereas PVE induced a decrease in tumorgrowth. In the nonoccluded part of the liver, bothPVE and PVL increased tumor volume. If con-firmed by further studies, especially in humans,these results highlight the negative side effects ofPVE and PVL on tumor growth and emphasize theneed for careful selection of patients beforeperforming either PVE or PVL before hepatec-tomy to avoid rapid development of LM in thefuture remnant liver.

REFERENCES

1. Nordlinger B, Sorbye H, Glimelius B, Poston GJ, Schlag PM,Rougier P, et al. Perioperative chemotherapy with FOL-FOX4 and surgery versus surgery alone for resectable livermetastases from colorectal cancer (EORTC Intergroup trial40983): a randomised controlled trial. Lancet 2008;371:1007-16.

2. Manfredi S, Lepage C, Hatem C, Coatmeur O, Faivre J,Bouvier AM. Epidemiology and management of livermetastases from colorectal cancer. Ann Surg 2006;244:254-9.

3. Garden OJ, Rees M, Poston GJ, Mirza D, Saunders M, Leder-mann J, et al. Guidelines for resection of colorectal cancerliver metastases. Gut 2006;55(Suppl 3):iii1-8.

4. Hobday TJ, Kugler JW, Mahoney MR, Sargent DJ, Sloan JA,Fitch TR, et al. Efficacy and quality-of-life data are related ina phase II trial of oral chemotherapy in previously un-treated patients with metastatic colorectal carcinoma.J Clin Oncol 2002;20:4574-80.

5. Makuuchi M, Thai BL, Takayasu K, Takayama T, Kosuge T,Gunven P, et al. Preoperative portal embolization to in-crease safety of major hepatectomy for hilar bile duct carci-noma: a preliminary report. Surgery 1990;107:521-7.

6. de Baere T, Roche A, Vavasseur D, Therasse E, Indushekar S,EliasD, et al. Portal vein embolization: utility for inducing lefthepatic lobe hypertrophy before surgery. Radiology 1993;188:73-7.

7. AzoulayD, CastaingD, Smail A, AdamR, Cailliez V, Laurent A,et al.Resection of nonresectable livermetastases fromcolorec-tal cancer after percutaneous portal vein embolization. AnnSurg 2000;231:480-6.

8. Honjo I, Suzuki T, Ozawa K, Takasan H, Kitamura O. Liga-tion of a branch of the portal vein for carcinoma of theliver. Am J Surg 1975;130:296-302.

9. Aussilhou B, Lesurtel M, Sauvanet A, Farges O, Dokmak S,Goasguen N, et al. Right portal vein ligation is as efficient asportal vein embolization to induce hypertrophy of the leftliver remnant. J Gastrointest Surg 2008;12:297-303.

10. Broering DC, Hillert C, Krupski G, Fischer L, Mueller L,Achilles EG, et al. Portal vein embolization vs. portal vein li-gation for induction of hypertrophy of the future liver rem-nant. J Gastrointest Surg 2002;6:905-13.

11. Denys AL, Abehsera M, Sauvanet A, Sibert A, Belghiti J,Menu Y. Failure of right portal vein ligation to induce leftlobe hypertrophy due to intrahepatic portoportal collat-erals: successful treatment with portal vein embolization.AJR Am J Roentgenol 1999;173:633-5.

12. Capussotti L, Muratore A, Baracchi F, Lelong B, Ferrero A,Regge D, et al. Portal vein ligation as an efficient method ofincreasing the future liver remnant volume in the surgicaltreatment of colorectal metastases. Arch Surg 2008;143:978-82.

13. PamechaV, LeveneA,Grillo F,WoodwardN,Dhillon A,David-sonBR.Effect of portal vein embolisationon thegrowth rateofcolorectal liver metastases. Br J Cancer 2009;100:617-22.

14. Hayashi S, Baba Y, Ueno K, Nakajo M, Kubo F, Ueno S, et al.Acceleration of primary liver tumor growth rate in embol-ized hepatic lobe after portal vein embolization. Acta Radiol2007;48:721-7.

15. BarbaroB,DiStasiC,NuzzoG,VelloneM,GiulianteF,MaranoP.Preoperative right portal vein embolization in patients withmetastatic liver disease. Metastatic liver volumes after RPVE.Acta Radiol 2003;44:98-102.

16. Elias D, De Baere T, Roche A, Mducreux Leclere J, Lasser P.During liver regeneration following right portal

SurgeryVolume 149, Number 4

Maggiori et al 503

embolization the growth rate of livermetastases ismore rapidthan that of the liver parenchyma. Br J Surg 1999;86:784-8.

17. Kokudo N, Tada K, Seki M, Ohta H, Azekura K, Ueno M,et al. Proliferative activity of intrahepatic colorectal metasta-ses after preoperative hemihepatic portal vein emboliza-tion. Hepatology 2001;34:267-72.

18. Onishi H, Murakami T, Kim T, Hori M, Iannaccone R,Kuwabara M, et al. Hepatic metastases: detection withmulti-detector row CT, SPIO-enhanced MR imaging, andboth techniques combined. Radiology 2006;239:131-8.

19. Martin F, Caignard A, Jeannin JF, Leclerc A, Martin M.Selection by trypsin of two sublines of rat colon cancer cellsforming progressive or regressive tumors. Int J Cancer 1983;32:623-7.

20. Robertson JH, Yang SY, Iga AM, Seifalian AM, Winslet MC.An in vivo rat model for early development of colorectalcancer metastasis to liver. Int J Exp Pathol 2008;89:447-57.

21. Wilms C, Mueller L, Lenk C, Wittkugel O, Helmke K, Krup-ski-Berdien G, et al. Comparative study of portal vein embo-lization versus portal vein ligation for induction ofhypertrophy of the future liver remnant using a mini-pigmodel. Ann Surg 2008;247:825-34.

22. Panis Y, Ribeiro J, Chretien Y, Nordlinger B. Dormant livermetastases: an experimental study. Br J Surg 1992;79:221-3.

23. Yamakado K, Takeda K,MatsumuraK,Nakatsuka A,HiranoT,Kato N, et al. Regeneration of the un-embolized liver paren-chyma following portal vein embolization. J Hepatol 1997;27:871-80.

24. Panis Y, McMullan DM, Emond JC. Progressive necrosis af-ter hepatectomy and the pathophysiology of liver failure af-ter massive resection. Surgery 1997;121:142-9.

25. Furrer K, Tian Y, Pfammatter T, Jochum W, El-Badry AM,Graf R, et al. Selective portal vein embolization and ligationtrigger different regenerative responses in the rat liver.Hepatology 2008;47:1615-23.

26. Goto Y, Nagino M, Nimura Y. Doppler estimation of portalblood flow after percutaneous transhepatic portal vein em-bolization. Ann Surg 1998;228:209-13.

27. Shimamura T, Nakajima Y, Une Y, Namieno T, Ogasawara K,Yamashita K, et al. Efficacy and safety of preoperative percu-taneous transhepatic portal embolization with absolute eth-anol: a clinical study. Surgery 1997;121:135-41.

28. Madoff DC, Abdalla EK, Vauthey JN. Portal vein emboliza-tion in preparation for major hepatic resection: evolution

of a new standard of care. J Vasc Interv Radiol 2005;16:779-90.

29. de Baere T, RocheA, Elias D, Lasser P, Lagrange C, Bousson V.Preoperative portal vein embolization for extension ofhepatectomy indications. Hepatology 1996;24:1386-91.

30. Heinrich S, Jochum W, Graf R, Clavien PA. Portal vein liga-tion and partial hepatectomy differentially influencegrowth of intrahepatic metastasis and liver regeneration inmice. J Hepatol 2006;45:35-42.

31. Yokoama Y, Wawrzyniak A, Sarmadi AM, Baveja R,Gruber HE, Clemens MG, et al. Hepatic arterial flow be-comes the primary supply of sinusoids following partialportal vein ligation in rats. J Gastroenterol Hepatol2006;21:1564-74.

32. Peeters CF, de Waal RM, Wobbes T, Westphal JR, RuersTJ. Outgrowth of human liver metastases after resectionof the primary colorectal tumor: a shift in the balance be-tween apoptosis and proliferation. Int J Cancer 2006;119:1249-53.

33. Yokoyama Y, Nagino M, Nimura Y. Mechanisms of hepa-tic regeneration following portal vein embolization andpartial hepatectomy: a review. World J Surg 2007;31:367-74.

34. Ribero D, Abdalla EK, Madoff DC, Donadon M, Loyer EM,Vauthey JN. Portal vein embolization before major hepatec-tomy and its effects on regeneration, resectability and out-come. Br J Surg 2007;94:1386-94.

35. Pamecha V, Glantzounis G, Davies N, Fusai G, Sharma D,Davidson B. Long-term survival and disease recurrencefollowing portal vein embolisation prior to major hepatec-tomy for colorectal metastases. Ann Surg Oncol 2009;16:1202-7.

36. AbulkhirA,LimongelliP,HealeyAJ,DamrahO,TaitP, JacksonJ,et al. Preoperative portal vein embolization formajor liver resec-tion: a meta-analysis. Ann Surg 2008;247:49-57.

37. KianmaneshR,FargesO,AbdallaEK,SauvanetA,RuszniewskiP,Belghiti J. Right portal vein ligation: a new planned two-step all-surgical approach for complete resectionof primary gastrointes-tinal tumors with multiple bilateral liver metastases. J Am CollSurg 2003;197:164-70.

38. Beal IK, Anthony S, Papadopoulou A, Hutchins R, Fusai G,Begent R, et al. Portal vein embolisation prior to hepatic re-section for colorectal liver metastases and the effects of peri-procedure chemotherapy. Br J Radiol 2006;79:473-8.