ORIGINAL ARTICLE

Expression of Matrix Metalloproteinase-9 InPredicting Prognosis of HepatocellularCarcinoma After Liver TransplantationDeniz Nart,1 Banu Yaman,4 Funda Yılmaz,1 Murat Zeytunlu,2 Zeki Karasu,3 and Murat Kılıc2

1Department of Pathology; 2Department of General Surgery, and 3Department of Gastroenterology,Faculty of Medicine, University of Ege, Izmir, Turkey; and 4Rize 82 Year State Hospital, Department ofPathology, Rize, Turkey

Matrix metalloproteinases (MMPs) are known to play an important role in cell migration during cancer invasion by degradingextracellular matrix proteins. This study aimed to determine the role of MMP-9 in hepatocellular carcinoma (HCC) carcino-genesis. Eighty-nine cases who underwent liver transplantation for HCC in cirrhotic liver were selected for this study. The tu-mor characteristics such as nodule number, maximal diameter, portal vein invasion, and the preoperative alpha-fetoproteinlevels were reviewed. The intensity of immunostaining and the percentage of immunoreactive cells with MMP-9 were eval-uated. All patients were evaluated for HCC recurrence and/or death, and cause of death was noted. There was a lower sur-vival and more recurrence risk among participants with 4 or more nodules exceeding 3 cm in diameter, with poorlydifferentiated tumor, and with large-vessel involvement. Eleven patients developed recurrent HCC (12.4%). Twelve patientsdied as a result of HCC (13.5%). Among 89 HCCs, the incidences of a weak (þ) and moderate (þþ) expression of MMP-9in carcinoma cells were 30.3% (23/89) and 43.8% (39/89), respectively. Increased expression and intensity of MMP-9 werefound to be inversely associated with poor tumor differentiation (P ¼ 0.016, P ¼ 0.009, respectively). A significant correla-tion between expression and intensity of MMP-9 and large vascular invasion (P ¼ 0.01, and P ¼ 0.03) was also observed.As far as prognosis is concerned, increased immunoreactivity and intensity of MMP-9 were found to exert an unfavorableimpact on overall survival rates (P < 0.01, P ¼ 0.01, respectively) and recurrences (P ¼ 0.001, P ¼ 0.02). Multivariate anal-yses revealed that MMP-9 staining percentage (P ¼ 0.007) and portal vein invasion (P ¼ 0.002) were independent predic-tors of survival, whereas the only independent predictor of recurrences was portal vein invasion (P ¼ 0.007). In this study,our results indicate a positive association between MMP-9 expression and histopathologic parameters that indicate poorprognosis. We conclude that together, MMP-9 staining percentage and portal vein invasion in HCC may aid to predict pooroutcome. Nevertheless MMP-9 staining percentage is expected to be a potential predictive marker on survival and needs tobe studied more in detail. Liver Transpl 16:621-630, 2010. VC 2010 AASLD.

Revised January 7, 2010; accepted January 11, 2010.

Hepatocellular carcinoma (HCC) is one of the mostcommon cancers worldwide, and still has an overallpoor prognosis.1 Chronic liver disease, especially dueto hepatitis B virus (HBV) and hepatitis C virus (HCV)infection complicated with cirrhosis, is an importantevoking factor for HCC development.2-4 Although liverresection is the first step in HCC treatment, livertransplantation (LT) is the only curative therapy inpatients with HCC and advanced concomitant cirrho-sis. Milan5 and the University of California, San Fran-

cisco (UCSF)6 criteria are the most widely used sys-tems to improve the survival in transplant patientsand are based on number of nodules, maximum di-ameter, or both. UCSF criteria, based on pretrans-plant imaging, also facilitate a discriminate prognosisafter LT.7 However, the evaluation of histopathologicalfeatures of the tumor in the explanted liver can givebetter information about tumor behavior.8-10 In addi-tion to the histopathological evaluation, molecularmarkers can provide additional information for the

Abbreviations: AFP, alpha-fetoprotein; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; LT, livertransplantation; MMP, matrix metalloproteinase.Address reprint requests to Banu Yaman, 8 Sokak No:25 K:3 D:8, Bornova-Izmir, Turkey. Telephone: þ90 506 5998539; FAX: þ90 2324614003; E-mail: drbanuyaman@yahoo.com

DOI 10.1002/lt.22028Published online in Wiley InterScience (www.interscience.wiley.com).

LIVER TRANSPLANTATION 16:621-630, 2010

VC 2010 American Association for the Study of Liver Diseases.

assessment of the tumor progression, invasion, andmetastasis. The ability of cancer cells to degrade theextracellular matrix and basement membrane is animportant initial step in the process of tumor inva-sion.11-13 Matrix metalloproteinases (MMPs), secretedby tumor cells, are extracellular matrix-degradingenzymes that enhance tumor invasiveness and metas-tasis.14,15 MMP-9 plays a key role in the invasive abil-ity and the degradation of the basement membrane,which is composed of type IV collagen.12 Previousstudies have shown that MMP-9 is predictive of amore invasive and metastatic character of HCC.16,17 Itwas also demonstrated in an in vitro study using HCCcell lines that invasion activity in hepatoma cells wasincreased by up-regulating MMP gene expression.18

The present study was carried out to gain furtherinsight into the immunohistochemical expression ofMMP-9 in HCC, in relation to the classical clinicopa-thological parameters and the clinical outcome.

PATIENTS AND METHODS

The study group comprised all 89 native hepatectomyspecimens containing HCC among consecutive ortho-topic LTs between 1997 and 2006. The pretransplan-tation selection criteria for patients with HCC werepreviously described by the Milan and the San Fran-cisco groups.3,12,19 The study group included 74 menand 15 women with an average age of 53.6 6 6.0years (range ¼ 37-66 years). The etiologies of underly-ing liver disease of the 89 patients with HCC are sum-marized in Table 1. Preoperative alpha-fetoprotein(AFP) levels were available in 62 of the patients. Allexplanted livers were evaluated for total number ofnodules, diameter of tumor nodules, and the presenceof macroinvasion of the portal vein. The specimenswere fixed in formalin and embedded in paraffin. Rou-tine histological examination was performed with he-matoxylin and eosin (HE) staining. HE-stained paraf-fin tissue sections were reviewed and analyzed for thetumor differentiation (well, moderate, poor) defined

TABLE 1. Etiologies of the Underlying Liver Disease

Etiology n (%)

HBV 69 (77.5)HCV 9 (10.1)HBV þ HDV 4 (4.5)Alcohol 3 (3.4)Cryptogenic 2 (2.3)Primary biliary cirrhosis 1 (1.1)HBV þ HCV 1 (1.1)

Total 89 (100)

Figure 1. MMP-9 immunostaining in nontumoral cirrhotic liver (A) in hepatocytes and inflammatory cells in portal tract(magnification, 340), (B) in hepatocytes (3200), and (C) in sinusoidal lining cells and in hepatocytes (3400).

622 NART ET AL.

LIVER TRANSPLANTATION.DOI 10.1002/lt. Published on behalf of the American Association for the Study of Liver Diseases

according to Edmondson criteria.20 Blood vessel inva-sion was also recorded.

Follow-up was available for all patients. Mean sur-vival time was 98 months (range ¼ 90-105 months).All patients were evaluated for HCC recurrence and/or death, and cause of death was noted.

Immunohistochemistry

Immunohistochemical staining was performed on 4-lm-thick formalin-fixed paraffin-embedded sections,using an avidin-biotin immunoperoxidase technique,after deparaffinization in xylene and rehydrationthrough graded alcohols. Microwave-mediated antigenretrieval in 10 mM (pH 6.0) citrate buffer for 15minutes and endogenous peroxidase activity using0.3% hydrogen peroxide were performed. Sectionswere incubated with monoclonal antibody MMP-9 (GE-213, 1:100 dilution; Neomarkers). A standard avidin-biotin peroxidase complex technique (Vectastain Elite;Vector Laboratories, Burlingame, CA) was used for vis-ualization, with diaminobenzidine as the chromogen.Finally, sections were counterstained with hematoxy-lin. Tonsil tissue was selected for positive controlaccording to the manufacturer’s recommendations.

Evaluation of Immunohistochemistry

Immunohistochemical evaluation was performed by 2pathologists who were blinded to the clinical data. Inthe nontumoral cirrhotic liver, MMP-9 staining wasgranular in hepatocytes and Kupffer cells; in portaltracts, MMP-9 was also weakly stained by inflamma-tory cells (Fig. 1). In tumor cells, the MMP-9

TABLE 2. Clinicopathological Characteristics

of the Patients

Clinicopathological Characteristics

N (%)Parameters Variable

Age 35-40 2 (2.2)41-50 22 (24.7)51-60 53 (59.6)61-66 12 (13.5)

Sex Male 74 (83.1)Female 15 (16.9)

Number of nodules 1 31 (34.8)2 15 (16.9)3 9 (10.1)

>4 34 (38.2)Diameter of tumor 1-�3 cm 61 (68.5)

>3 cm 28 (31.5)Differentiation Well 20 (22.5)

Moderate 52 (58.4)Poor 17 (19.1)

Portal veininvolvement

No 47 (52.8)

Segmental vascularinvasion

36 (40.4)

Involvement of largevessels

6 (6.7)

Recurrence Yes 11 (12.4)No 78 (87.6)

Final outcome Alive 75 (84.3)Death (due to

carcinoma)12 (13.5)

Death (due toother causes)

2 (2.2)

TABLE 3. Univariate Analyses of Recurrence and Survival with Clinicopathological Characteristics

Parameters

Recurrence Survival

n No-n (%) Yes-n (%) P Value Ex ca-n (%) Ex other-n (%) Alive-n (%) P Value

Sex 0.43 0.85Male 74 64 (86.5) 10 (13.5) 10 (13.5) 2 (2.7) 62 (83.8)Female 15 14 (93.3) 1 (6.7) 2 (13.3) 0 13 (86.7)

Number of nodules 0.001 0.004

�3 55 53 (96.4) 2 (3.6) 3 (5.5) 1 (1.8) 51 (92.7)�4 34 25 (73.5) 9 (26.5) 9 (26.5) 1 (2.9) 24 (70.6)

Diameter of tumor 0.0001 0.005

1-�3 cm 61 59 (96.7) 2 (3.3) 4 (6.6) 2 (3.3) 55 (90.1)>3 cm 28 19 (67.9) 9 (39.1) 8 (28.6) 0 20 (71.4)

Differentiation 0.0001 0.001

Well 20 20 (100) 0 2 (10) 0 18 (90)Moderate 52 49 (94.2) 3 (5.8) 1 (1.9) 2 (3.9) 49 (94.2)Poor 17 9 (52.9) 8 (47.1) 9 (52.9) 0 8 (47.1)

Portal vein involvement 0.0001 0.0001

No 47 47 (100) 0 2 (4.3) 1 (2.1) 44 (93.6)Segmental vascular invasion 36 30 (83.3) 6 (16.7) 4 (11.1) 1 (2.8) 31 (86.1)Involvement of large vessels 6 1 (16.7) 5 (83.3) 6 (100) 0 0

EXPRESSION OF MMP-9 IN HEPATOCELLULAR CARCINOMA 623

LIVER TRANSPLANTATION.DOI 10.1002/lt. Published on behalf of the American Association for the Study of Liver Diseases

staining pattern was cytoplasmic, membranous, orcytoplasmic and membranous. Staining intensity andthe percentage of immunoreactive cells over total tu-mor cells were taken into consideration throughoutthe evaluation process. No interobserver variabilitywas accepted. The intensity was scored as; þ ¼ 1(weak), þþ ¼ 2 (moderate), and þþþ ¼ 3 (strong),respectively. A score of 0 was given if �5% of cellswere stained, score 1 if 6%-25% of cells were immu-noreactive, score 2 if 26%-50% of cells were immuno-reactive, score 3 if 51%-75% of cells were immunore-active, and score 4 if �76% of cells wereimmunoreactive.

Statistical Analysis

The significance of the relationship between expres-sion of MMP-9 and clinicopathological parameterswas evaluated with univariate analysis using Pearsonchi-squared test and Fisher’s exact probability test.The univariate (log-rank, Mantel-Cox) tests, Kaplan-Meier survival curves, and the multivariate analysis

were used for the effect of MMP-9 differential expres-sion on postoperative survival rates and recurrences.All statistical analyses were performed using the Sta-tistical Package for the Social Sciences, version 13.0for Windows (SPSS Inc., Chicago, IL). The P values of0.05 or less were considered as statisticallysignificant.

RESULTS

Patients were predominantly men (74 men, 15women) with an average age of 53.6 6 6.0 years.Underlying liver diseases in these 89 patients areindicated in Table 1. The mean preoperative AFPvalue was 291.86 IU/mL (range ¼ 1.21-10,558 IU/mL). The clinicopathological characteristics of our se-ries are shown in Table 2. Eleven patients developedrecurrent HCC (12.4%). Twelve patients died as aresult of HCC (13.5%). The distribution of parameterswith respect to recurrence and survival are shown inTable 3.

Figure 2. Schematic representation of the association between the number of nodules (A,B) and diameter of tumor (C,D) andthe overall survival and recurrence time.

624 NART ET AL.

LIVER TRANSPLANTATION.DOI 10.1002/lt. Published on behalf of the American Association for the Study of Liver Diseases

Among 55 patients with 3 or fewer nodules, 2 devel-oped recurrence (3.6%) and 3 of them died as a resultof HCC (5.5%), whereas 9 (26.5%) of 34 patients with4 or more nodules had recurrence and died as aresult of HCC. The mean survival time in those with 3or fewer nodules was 105.78 6 3.5 months, whereasin those with 4 or more nodules, survival durationwas 81.24 6 8.7 months. There was a lower survivaland more recurrence risk among participants with 4

or more nodules compared to those with 3 or fewernodules (P ¼ 0.004, and P ¼ 0.001, respectively; log-rank test). Considering the size of the largest nodule,having a diameter of 3 cm or less was associated withbetter survival time (104.56 6 3.6 months) and recur-rence risk (P ¼ 0.005, and P ¼ 0.0001, log-rank test)compared to the nodules having a diameter of morethan 3 cm (67.03 6 7.4 months) (Fig. 2).

The patients with poorly differentiated tumors hadmore recurrences (47.1%) than the other patients (0%for well-differentiated and 5.8% for moderately differ-entiated) (P ¼ 0.0001, log-rank test). Only 2 patients(10%) with well-differentiated tumor (mean survivaltime 100.02 6 7.9 months) died because of disease,whereas 9 patients (52.9%) died due to poorly differ-entiated tumor (mean survival time ¼ 48.86 6 4.2months) (P ¼ 0.001, log-rank test). All of the patientswith large vessel involvement died because of disease,and the majority of this group (83.3%) had recur-rence. There was a statistically significant differencebetween survival and recurrence of the group withlarge vessel involvement when compared to the others(P < 0.0001, log-rank test) (Fig. 3). The mean survivaltime with no vessel involvement was 107.03 6 3.4months, whereas mean survival time with large vesselinvolvement was 15.50 6 5.8 months.

Among the 62 patients having preoperative AFP lev-els, 5 patients developed recurrence (mean ¼ 2189.98IU/mL) whereas 57 patients did not develop recur-rence (mean ¼ 125.35 IU/mL). A total of 57 of 62patients survived (mean ¼ 131.56 IU/mL) while 5 ofthem died due to recurrence of the tumor (mean ¼2119.30 IU/mL). Neither recurrence (P ¼ 0.12) norsurvival (P ¼ 0.69) showed statistically significant cor-relation with the preoperative AFP level.

MMP-9 was mainly expressed within the cytoplasmand cytoplasmic membranes of HCC tumor cells (Fig. 4).Among 89 HCC cases, the incidence of a positive expres-sion of MMP-9 in tumor cells was 74.2% (66/89), withthe incidence of strong immunoreactivity of 18% (16/89). The association between the clinicopathological pa-rameters and MMP-9 expression and intensity is dem-onstrated in Table 4. There were no significant correla-tions amid the expression pattern, intensity, andpercentage of MMP-9 staining or number of nodules andsize of tumor (P > 0.05). Increased expression and inten-sity of MMP-9 were found to be inversely associated withtumor differentiation (P¼ 0.016, P¼ 0.009, respectively)(Fig. 5). Cytoplasmic and membranous expression ofMMP-9 correlated with decreased differentiation (P ¼0.001). Cytoplasmic or membranous locations of MMP-9 were more often observed in well-differentiated ormoderately differentiated tumors.

A significant correlation was demonstrated betweenincreased MMP-9 expression and intensity and largevessel involvement (P ¼ 0.01 and P ¼ 0.03,respectively).

AFP levels were 321, 57, 33, 62, and 5291 for MMPstaining percentage of 0%-5%, 6%-25%, 26%-50%,51%-75%, and 76%-100%, respectively (P ¼ 0.408).The mean AFP levels were 321, 72, 15, and 2169 for

Figure 3. The graphic of the impact of (A,B) thedifferentiation grade and (C) portal vein involvement on theoverall survival and recurrence time.

EXPRESSION OF MMP-9 IN HEPATOCELLULAR CARCINOMA 625

LIVER TRANSPLANTATION.DOI 10.1002/lt. Published on behalf of the American Association for the Study of Liver Diseases

Figure 4. (A) Cytoplasmic (magnification, 340), (B) membranous (3100), and (C,D) cytoplasmic and membranous (3200,3400, respectively) strong immunoreactivity of MMP-9 in HCC cells.

TABLE 4. Correlation of Matrix Metalloproteinase 9 Expression and Intensity with Clinicopathological Parameters

Parameters

MMP-9 Staining Percentage MMP-9 Staining Intensity

0%-

5%

n (%)

6%-

25%

n (%)

26%-

50%

n (%)

51%-

75%

n (%)

76%-

100%

n (%)

P

Value

0

n (%)

þn (%)

þþn (%)

þþþn (%)

P

Value

Number ofnodules

>0.05 >0.05

� 3 16 (69.6) 18 (60) 13 (65) 6 (50) 2 (50) 16 (69.6) 20 (74.1) 14 (46.7) 5 (55.6)� 4 7 (30.4) 12 (40) 7 (35) 6 (50) 2 (50) 7 (30.4) 7 (25.9) 16 (53.3) 4 (44.4)

Diameter oftumor

>0.05 >0.05

1-�3 cm 18 (78.2) 21 (70) 15 (75) 6 (50) 1 (25) 18 (78.2) 19 (70.4) 21 (70) 3 (33.3)>3 cm 5 (21.8) 9 (30) 5 (25) 6 (50) 3 (75) 5 (21.8) 8 (29.6) 9 (30) 6 (66.7)

Differentiation 0.016 0.009

Well 7 (30.4) 7 (23.3) 5 (25) 1 (8.3) 0 7 (30.4) 7 (25.9) 6 (20) 0Moderate 14 (60.9) 20 (66.7) 12 (60) 4 (33.0) 2 (50) 14 (60.9) 19 (70.4) 15 (50) 4 (44.4)Poor 2 (8.7) 3 (10) 3 (15) 7 (58.3) 2 (50) 2 (8.7) 1 (3.7) 9 (30) 5 (55.6)

Portal veininvolvement

0.01 0.03

No 16 (69.6) 13 (43.3) 13 (65) 5 (41.7) 0 16 (69.6) 17 (63) 13 (43.3) 1 (11.1)Segmental

vascularinvasion

7 (30.4) 17 (56.7) 5 (25) 5 (41.7) 2 (50) 7 (30.4) 10 (37) 14 (46.7) 5 (55.6)

Involvement oflarge vessels

0 0 2 (10) 2 (16.7) 2 (50) 0 0 3 (10) 3 (33.3)

Total 23 30 20 12 4 23 27 30 9

626 NART ET AL.

LIVER TRANSPLANTATION.DOI 10.1002/lt. Published on behalf of the American Association for the Study of Liver Diseases

the MMP staining intensity of 0, (þ), (þþ), and (þþþ),respectively (P ¼ 0.301).

Multivariate analysis was employed to elucidate amore precise role of differentiation, portal vein inva-sion, survival, and recurrence as independent factorsfor MMP-9 staining. The results showed that MMP-9staining percentage was mostly affected by differentia-tion (P ¼ 0.001), whereas MMP-9 staining intensitywas mostly affected by portal vein invasion (P <0.0001).

With respect to prognosis, increased immunoreactiv-ity and intensity of MMP-9 were found to exert an unfav-orable impact on overall survival results (P < 0.01 andP ¼ 0.01, respectively). The mean survival time ofpatients with 76%-100% MMP-9 expression was 16.6 64.2 months, whereas mean survival time was 84.26 3.7months for patients with 0%-5% expression. Patientswith greater MMP-9 expression (more than 51%) and in-tensity (þþþ) showed more recurrences than the others(P ¼ 0.001 and P ¼ 0.02; Table 5).

On multivariate analysis, survival was mostly affectedby portal vein invasion (P ¼ 0.002) and MMP-9 stainingpercentage (P ¼ 0.007) rather than the number, size, dif-ferentiation, and MMP staining intensity of nodules,while recurrence was mostly affected only by portal veininvasion (P ¼ 0.007). MMP-9 staining percentage and in-

tensity did not correlate to the recurrence; however, wefound the 95% confidence interval very wide on bothmultivariate analyses for survival and recurrencesbecause we studied a broad patient group (Table 6).

DISCUSSION

Hepatocellular carcinoma is one of the most commoncancers worldwide.1 The major risk factor is liver cir-rhosis associated with chronic HBV and HCV infec-tions, aflatoxin B exposure, and various metabolicdisorders.21-24 In Turkey, most patients with HCC ex-hibit concomitant cirrhosis caused by HBV andHCV.23 Eighty percent of HCCs are encapsulated byfibrous tissue, and almost all HCCs are associatedwith an extracellular matrix-rich diseased liver.17

In this work, the etiology of the underlying liver dis-ease, the number and the size of tumor nodules, thetumor differentiation degree, and the presence of por-tal vein involvement were evaluated and prognosticvalue was assessed by survival free from tumorrecurrence.

In our study, the number of patients with HCC whowere also infected with HBV and/or HCV was alsovery high. The number of nodules, the tumor size, thegrade of differentiation, and the presence of portal

TABLE 5. Univariate Analyses of Recurrence and Survival with Matrix Metalloproteinase 9 Expression and Intensity

Parameters

Recurrence Survival

n No-n (%) Yes-n (%) P Value Ex ca-n (%) Ex other-n (%) Alive-n (%) P Value

MMP-9 Staining percentage 0.001 <0.010%-5% 23 21 (91.3) 2 (8.7) 1 (4.4) 0 22 (95.6)6%-25% 30 28 (93.3) 2 (6.7) 1 (3.3) 0 29 (96.7)26%-50% 20 19 (95) 1 (5) 3 (15) 1 (5) 16 (80)51%-75% 12 9 (75) 3 (25) 4 (33.4) 1 (8.3) 7 (58.3)76%-100% 4 1 (25) 3 (75) 3 (75) 0 1 (75)

MMP-9 Staining Intensity 0.02 0.010 23 21 (91.3) 2 (8.7) 1 (4.4) 0 22 (95.6)þ 27 25 (92.6) 2 (7.4) 2 (7.4) 0 25 (92.6)þþ 30 27 (90) 3 (10) 5 (16.6) 2 (6.7) 23 (76.7)þþþ 9 5 (55.6) 4 (44.4) 4 (44.4) 0 5 (55.6)

Figure 5. Association betweenmatrix metalloproteinase 9expression and intensity anddifferentiation grade of the tumor.

EXPRESSION OF MMP-9 IN HEPATOCELLULAR CARCINOMA 627

LIVER TRANSPLANTATION.DOI 10.1002/lt. Published on behalf of the American Association for the Study of Liver Diseases

vein involvement were discovered as the factors thatinfluence the recurrence and prognosis. Our resultsare consistent with those previously reported by someother studies. In our study, preoperative AFP levelsdid not affect survival and recurrence. The limitednumber of available patients having preoperative AFPlevels and the extensive differences among the AFPlevels may have been statistically insignificant.

Patients with HCC have a poor prognosis because ofearly recurrence, and the molecular mechanisms thatunderlie the clinical behavior of HCC are still not wellunderstood.22,23,25,26

Metastasis is a complex cascade of events, the firststage of which is invasion.11,25,27 In this process, thedestruction of the extracellular matrix, including thebasement membrane, is an essential initial step.28 Theloss of basement membrane integrity in HCC may alsocontribute to the development of early metastasis.25,29

Proteolytic enzymes, collectively known as matrixproteinases, are endogenous peptidases25,30,31 andhave been shown to play an important role in theproteolysis of various components of the extracellu-lar matrix.17 Matrix proteinases are also involved intissue damage during the inflammatory process bydegrading cellular and extracellular components.30

MMPs are 1 of the 4 subgroups of matrix protein-ases. MMP expression has been noted in severalhuman cancers, including breast, colon, ovary, andpancreatic cancers. MMPs are released from cells ininactive forms and must be activated by proteolyticenzymes (MMP activators), such as cathepsins andtrypsin, to exert the proteolytic activity.30,32-35

Enhanced MMP expression has been reported in vari-ous human malignant tumors.22

Studies involving HCC cell lines demonstrate thatmost HCC cell lines secrete increased levels ofMMPs.25 By secreting MMPs, HCC degrades andinvades these barriers of extracellular matrix andblood vessels, which eventually leads to its spread todistant organs as well as in the liver itself during dis-ease progression.33

There are currently at least 20 MMPs, and they canbe subclassified according to their substrates: collage-nases (MMP-1, MMP-8, MMP-13), gelatinases (MMP-2,MMP-9), stromelysins (MMP-3, MMP-7, MMP-10), andmembrane-type (MT-MMP-1, MT-MMP-2, MT-MMP-3), which are bound to epithelial cell membranes and

can activate other MMPs. The metalloproteinases thatdegrade type IV collagen, which is the principal com-ponent of the basement membrane, are MMP-2, MMP-7, and MMP-9.17,25 The cells producing these enzymesinclude not only connective tissue but also tumorcells.17

MMP-9 (a 92-kDa gelatinase type IV collagenase[gelatinase B]) plays an important role in tumor inva-sion and angiogenesis.26,27,32 Secretion has beenreported in human hepatoma and glioblastoma cellsand in various cancer types including lung, colon,gastric, and breast cancer.18,17,27,32 Expressions ofMMP-9 in tumor tissues were of prognostic signifi-cance for poor overall survival of patients with gastriccarcinoma, HCC, and pancreatic adenocarcinoma.27

This study was carried out to gain further insightinto the immunohistochemical expression of MMP-9in HCC in relation to the classical clinicopathologicalparameters and the clinical outcome.

Previous studies have shown that MMP-9 expres-sion is a significant predictor of invasiveness, recur-rence, and metastasis of HCC.16,17,26,36-38 Regardingthe correlation of MMP messenger RNA and histopath-ological findings, the tumors with capsular infiltrationshowed a stronger expression of MMP-9 messengerRNA than those without capsular infiltration.17 More-over, plasma MMP-9 levels were significantly elevatedin the patients with HCC with macroscopic portal veininvasions compared to the patients with HCC withoutthose invasions.16,39 However, one study reported thatMMP-9 polymorphisms had no influence on the riskof recurrent HCC after LT.40

In this study, we were able to document progressiveexpression of MMP-9 as HCC progresses from low tohigh grade. There is a gradual increased expression,which further correlates with the number of nodules,the tumor size, the tumor differentiation, and the vas-cular invasion. Multivariate analysis showed that dif-ferentiation mostly affected MMP-9 staining percent-age while portal vein invasion affected MMP-9staining intensity. MMP-9 expression was significantlycorrelated with portal vein invasion and with undiffer-entiated tumors which may reflect the biologicalaggressive behavior of the tumor. With respect toprognosis, increased immunoreactivity and intensityof MMP-9 were found to exert an unfavorable impacton overall survival. The observation of increased

TABLE 6. Multivariate Analysis of Each Individual Parameter on Survival and Recurrence

Parameters

Survival Recurrence

Hazard Ratio 95% CI P Value Hazard Ratio 95% CI P Value

Number of nodules 1.44 0.68-3.05 0.33 1.27 0.45-3.58 0.64Diameter of tumor 0.97 0.57-1.67 0.94 1.49 0.81-2.74 0.19Differentiation 1.52 0.30-7.67 0.60 5.44 0.72-40.83 0.099Portal vein invasion 12.78 2.59-63.08 0.002 31.56 2.56-388.47 0.007

MMP-9 staining percentage 11.84 1.99-70.42 0.007 9.20 0.92-92.13 0.059MMP-9 staining intensity 0.10 0.01-1.02 0.052 0.05 0.01-1.05 0.054

628 NART ET AL.

LIVER TRANSPLANTATION.DOI 10.1002/lt. Published on behalf of the American Association for the Study of Liver Diseases

expression of MMP-9 further emphasizes the role ofthis molecule in HCC carcinogenesis and its relationto poor prognostic parameters and survival.

Although recurrence was affected only by portalvein invasion, survival was affected by both portalvein invasion and MMP-9 staining percentage. Wealso demonstrated that MMP-9 staining percentagedirectly affects survival.

It should be noted that there were some limitationsin this study. First, our study population was mainlycomposed of patients with HBV-induced HCC. Itwould therefore be important to confirm these find-ings in patients with other liver diseases. In addition,the sample size was relatively small. The study shouldtherefore be viewed as a hypothesis-generating andpreliminary work to be followed by larger prospectiveand multiethnic studies to confirm related consequentfindings. Despite these limitations, the present datasuggest that MMP-9 expression can be a candidate asa novel marker for predicting HCC behavior and canalso serve as a potential target in terms of preventionand treatment of the tumor.

ACKNOWLEDGMENTThe authors would like to thank Dr. Neslihan Incili forsampling the data of the patients.

REFERENCES

1. Bosch FX, Ribes J, Dıaz M, Cleries R. Epidemiology ofhepatocellular carcinoma. Clin Liver Dis 2005;9:191-211.

2. Schafer DF, Sorrell MF. Hepatocellular carcinoma. Lan-cet 1999;353:1253-1257.

3. Tsukuma H, Hiyama T, Tanaka S, Nakao M, Yabuuchi T,Kitamura T, et al. Risk factors for hepatocellular carci-noma among patients with chronic liver disease. N EnglJ Med 1993;328:1797-1801.

4. Koike Y, Shiratori Y, Sato S, Obi S, Teratani T, ImamuraM, et al. Risk factors for recurring hepatocellular carci-noma differ according to infected hepatitis virus-an anal-ysis of 236 consecutive patients with a single lesion.Hepatology 2000;32:1216-1223.

5. Mazzaferro V, Regalia E, Doci R, Andreola S, PulvirentiA, Bozzetti F, et al. Liver transplantation for the treat-ment of small hepatocellular carcinomas in patients withcirrhosis. N Engl J Med 1996;334:693-699.

6. Yao FY, Ferrell L, Bass NM, Watson JJ, Bacchetti P,Venook A, et al. Liver transplantation for hepatocellularcarcinoma: expansion of the tumor size limits does notadversely impact survival. Hepatology 2001;33:1394-1403.

7. Yao FY, Xiao L, Bass NM, Kerlan R, Ascher NL, RobertsJP. Liver transplantation for hepatocellular carcinoma:validation of the UCSF-expanded criteria based on pre-operative imaging. Am J Transplant 2007;7:2587-2596.

8. Pawlik TM, Delman KA, Vauthey JN, Nagorney DM, NgIO, Ikai I, et al. Tumor size predicts vascular invasionand histologic grade: implications for selection of surgi-cal treatment for hepatocellular carcinoma. Liver Transpl2005;11:1086-1092.

9. Parfitt JR, Marotta P, Alghamdi M, Wall W, Khakhar A,Suskin NG, et al. Recurrent hepatocellular carcinoma af-ter transplantation: use of a pathological score on

explanted livers to predict recurrence. Liver Transpl2007;13:543-551.

10. Llovet JM, Bru C, Bruix J. Prognosis of hepatocellularcarcinoma: the BCLC staging classification. Semin LiverDis 1999;19:329-338.

11. Liotta LA. Tumor invasion and metastases--role of theextracellular matrix: Rhoads Memorial Award lecture.Cancer Res 1986;46:1-7.

12. Liotta LA, Kohn EC. The microenvironment of thetumour-host interface. Nature 2001;411:375-379.

13. Comoglio PM, Trusolino L. Invasive growth: from devel-opment to metastasis. J Clin Invest 2002;109:857-862.

14. Coussens LM, Werb Z. Matrix metalloproteinases andthe development of cancer. Chem Biol 1996;3:895-904.

15. Littlepage LE, Egeblad M, Werb Z. Coevolution of cancerand stromal cellular responses. Cancer Cell 2005;7:499-500.

16. Hayasaka A, Suzuki N, Fujimoto N, Iwama S, FukuyamaE, Kanda Y, et al. Elevated plasma levels of matrix metallo-proteinase-9 (92-kd type IV collagenase/gelatinase B) inhepatocellular carcinoma. Hepatology 1996;24:1058-1062.

17. Arii S, Mise M, Harada T, Furutani M, Ishigami S,Niwano M, et al. Overexpression of matrix metalloprotei-nase 9 gene in hepatocellular carcinoma with invasivepotential. Hepatology 1996;24:316-322.

18. Miyoshi A, Kitajima Y, Sumi K, Sato K, Hagiwara A, KogaY, et al. Snail and SIP1 increase cancer invasion byupregulating MMP family in hepatocellular carcinomacells. Br J Cancer 2004;90:1265-1273.

19. Nart D, Arikan C, Akyildiz M, Yuce G, Demirpolat G, Zey-tunlu M, et al. Hepatocellular carcinoma in liver trans-plant era: a clinicopathologic analysis. Transplant Proc2003;35:2986-2990.

20. Edmondson HA, Steıner PE. Primary carcinoma of theliver: a study of 100 cases among 48,900 necropsies.Cancer 1954;7:462-503.

21. Chen CJ, Wang LY, Lu SN, Wu MH, You SL, Zhang YJ,et al. Elevated aflatoxin exposure and increased risk ofhepatocellular carcinoma. Hepatology 1996;24:38-42.

22. Gao ZH, Tretiakova MS, Liu WH, Gong C, Farris PD, HartJ. Association of E-cadherin, matrix metalloproteinases,and tissue inhibitors of metalloproteinases with the pro-gression and metastasis of hepatocellular carcinoma. ModPathol 2006;19:533-540.

23. Karakayali H, Sevmis S, Moray G. Liver transplantationin patients with hepatocellular carcinoma: one center’sexperience. Transplant Proc 2008;40:213-218.

24. Cormier JN, Thomas KT, Chari RS, Pinson CW. Manage-ment of hepatocellular carcinoma. J Gastrointest Surg2006;10:761-780.

25. McKenna GJ, Chen Y, Smith RM, Meneghetti A, Ong C,McMaster R, et al. A role for matrix metalloproteinasesand tumor host interaction in hepatocellular carcinomas.Am J Surg 2002;183:588-594.

26. Jiang YF, Yang ZH, Hu JQ. Recurrence or metastasis ofHCC: predictors, early detection and experimental anti-angiogenic therapy. World J Gastroenterol 2000;6:61-65.

27. Zhang S, Li L, Lin JY, Lin H. Imbalance between expres-sion of matrix metalloproteinase-9 and tissue inhibitor ofmetalloproteinase-1 in invasiveness and metastasis ofhuman gastric carcinoma. World J Gastroenterol 2003;9:899-904.

28. Chung TW, Moon SK, Lee YC, Kim JG, Ko JH, Kim CH.Enhanced expression of matrix metalloproteinase-9 byhepatitis B virus infection in liver cells. Arch BiochemBiophys 2002;408:147-154.

29. Theret N, Musso O, L’Helgoualc’h A, Campion JP, Clem-ent B. Differential expression and origin of membrane-type 1 and 2 matrix metalloproteinases (MT-MMPs) in

EXPRESSION OF MMP-9 IN HEPATOCELLULAR CARCINOMA 629

LIVER TRANSPLANTATION.DOI 10.1002/lt. Published on behalf of the American Association for the Study of Liver Diseases

association with MMP2 activation in injured human liv-ers. Am J Pathol 1998;153:945-954.

30. Terada T, Okada Y, Nakanuma Y. Expression of immu-noreactive matrix metalloproteinases and tissue inhibi-tors of matrix metalloproteinases in human normal liversand primary liver tumors. Hepatology 1996;23:1341-1244.

31. Birkedal-Hansen H, Moore WG, Bodden MK, Windsor LJ,Birkedal-Hansen B, DeCarlo A, et al. Matrix metallopro-teinases: a review. Crit Rev Oral Biol Med 1993;4:197-250.

32. Roeb E, Bosserhoff AK, Hamacher S, Jansen B, DahmenJ, Wagner S, et al. Enhanced migration of tissue inhibi-tor of metalloproteinase overexpressing hepatoma cells isattributed to gelatinases: relevance to intracellular sig-naling pathways. World J Gastroenterol 2005;11:1096-1104.

33. Matsumoto E, Nakatsukasa H, Nouso K, Kobayashi Y,Nakamura S, Suzuki M, et al. Increased levels of tissueinhibitor of metalloproteinase-1 in human hepatocellularcarcinoma. Liver Int 2004;24:379-383.

34. Nuovo GJ, MacConnell PB, Simsir A, Valea F, FrenchDL. Correlation of the in situ detection of polymerasechain reaction-amplified metalloproteinase complemen-tary DNAs and their inhibitors with prognosis in cervicalcarcinoma. Cancer Res 1995;55:267-275.

35. Stamenkovic I. Matrix metalloproteinases in tumor inva-sion and metastasis. Semin Cancer Biol 2000;10:415-433.

36. Zhang Q, Chen X, Zhou J, Zhang L, Zhao Q, Chen G,et al. CD147, MMP-2, MMP-9 and MVD-CD34 are signifi-cant predictors of recurrence after liver transplantationin hepatocellular carcinoma patients. Cancer Biol Ther2006;5:808-814.

37. Zhong C, Guo RP, Shi M, Wei W, Yu WS, Li JQ. Expres-sion and clinical significance of VEGF and MMP-9 in he-patocellular carcinoma. Ai Zheng 2006;25:599-603.

38. Sun MH, Han XC, Jia MK, Jiang WD, Wang M, Zhang H,et al. Expressions of inducible nitric oxide synthase andmatrix metalloproteinase-9 and their effects on angio-genesis and progression of hepatocellular carcinoma.World J Gastroenterol 2005;11:5931-5937.

39. Maatta M, Soini Y, Liakka A, Autio-Harmainen H. Differ-ential expression of matrix metalloproteinase (MMP)-2,MMP-9, and membrane type 1-MMP in hepatocellularand pancreatic adenocarcinoma: implications for tumorprogression and clinical prognosis. Clin Cancer Res2000;6:2726-2734.

40. Wu LM, Zhang F, Xie HY, Xu X, Chen QX, Yin SY, et al.MMP2 promoter polymorphism (C-1306T) and risk of re-currence in patients with hepatocellular carcinoma aftertransplantation. Clin Genet 2008;73:273-278.

630 NART ET AL.

LIVER TRANSPLANTATION.DOI 10.1002/lt. Published on behalf of the American Association for the Study of Liver Diseases