Vol. 107 No. 5 May 2009

ORAL AND MAXILLOFACIAL PATHOLOGY Editor: Mark W. Lingen

Prognostic significance of CXCR-4 expression in oral squamouscell carcinomaJae-Il Lee, DDS, PhD,a Bo-Hyong Jin, DDS, PhD,b Mi-Ae Kim, DDS, MS,a

Hye-Jung Yoon, DDS, MS,a Sam-Pyo Hong, DDS, PhD,a and Seong-Doo Hong, DDS, PhD,a

Seoul, KoreaSCHOOL OF DENTISTRY AND DENTAL RESEARCH INSTITUTE, SEOUL NATIONAL UNIVERSITY

Objective. We investigated the prognostic significance of CXC chemokine receptor CXCR-4 expression in patientswith oral squamous cell carcinoma (OSCC) and its relationship with matrix metalloproteinase 2 (MMP-2), MMP-9, andKi-67 expression.Study design. The CXCR-4, MMP-2, MMP-9, and Ki-67 expression was assessed immunohistochemically in 74 OSCCpatients. The results were analyzed in connection with clinicopathologic factors.Results. The CXCR-4 expression was positive in 45 cases and significantly correlated with lymph node metastasis (P �.037), MMP-9 expression (P � .025), and Ki-67 expression (P � .001). Univariate analysis showed that CXCR-4expression, MMP-9 expression, Ki-67 expression, tumor size, lymph node metastasis, clinical stage, and recurrencepositively correlated with prognosis. Multivariate analysis indicated that CXCR-4 expression was an independentprognostic factor for poor survival in patients with OSCC.Conclusion. Expression of CXCR-4 is a significant prognostic indicator for poor survival in patients with OSCC andcorrelates with expression of MMP-9 and Ki-67. The inhibition of CXCR-4 represents a possible molecular approach to

the treatment of OSCC. (Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2009;107:678-684)

Oral squamous cell carcinoma (OSCC), the mostcommon malignant neoplasm in the oral cavity, is asignificant global public health threat.1 The manage-ment of OSCC varies considerably, but standardprocedure includes surgery with or without neckdissection, followed by adjuvant radiotherapy.2 De-spite all efforts and therapeutic developments, the5-year survival rate for head and neck cancers hasnot remarkably improved over the last 2 decades.1,3

In clinical practice, treatment planning and prognosisfor patients with OSCC are mainly based on theTNM classification.4 At the same time, clinical out-come does not always follow the predictions of those

The first two authors contributed equally to this study.aDepartment of Oral Pathology.bDepartment of Preventive Dentistry.Received for publication Jul 21, 2008; returned for revision Dec 18,2008; accepted for publication Dec 18, 2008.1079-2104/$ - see front matter© 2009 Published by Mosby, Inc.

doi:10.1016/j.tripleo.2008.12.047

678

parameters, suggesting that other factors related tothe patient or the biologic characteristics of the tu-mor may be relevant.

Chemokines are a superfamily of small cytokineswith the ability to chemoattract cells to target tissues.Recent data indicate that chemokine receptors maydirect lymphatic and hematogenous spread and addi-tionally influence the location of metastatic tumorgrowth.5 Expression of CXC chemokine receptorCXCR4 is involved in the lymph node or distant me-tastasis of several types of cancer,6-12 including oralcancer.13,14 Even so, there have been few reports aboutthe relationship of CXCR-4 expression with survival inOSCC patients.

Matrix metalloproteinases (MMPs), a family of zinc-dependent proteinases, are necessary for the degrada-tion of extracellular matrix and can be produced byinvasive tumor cells.15 MMP-2 and MMP-9 are closelyassociated with malignant potential in OSCCs.16,17 Re-cent studies suggest that the CXCR-4–chemokine li-

gand 12 (CXCL-12) interaction increases invasiveness

OOOOEVolume 107, Number 5 Lee et al. 679

through the up-regulation of MMP-2 or MMP-9 insome cancer cells.18-21

The Ki-67 antibody is reactive with human nuclearproteins associated with cell proliferation.22 Re-cently, CXCR-4 has been shown to directly or indi-

Table I. Correlation of chemokine receptor 4 (CXCR-

Variable n

CXCR-4–pos

n (%)

Age�60 yrs 37 22 (59.5%)�60 yrs 37 23 (62.2%)

GenderMale 56 33 (58.9%)Female 18 12 (66.7%)

DifferentiationWell 52 32 (61.5%)Mod./poorly 22 13 (59.1%)

Tumor sizeT1/T2 44 23 (52.3%)T3/T4 30 22 (73.3%)

LN metastasisPositive 26 20 (76.9%)Negative 48 25 (52.1%)

Clinical stageI/II 36 18 (50.0%)III/IV 38 27 (71.1%)

RecurrenceYes 25 17 (68.0%)No 49 28 (57.1%)

LI, Labeling index; LN, lymph node.

Table II. Correlation of matrix metalloproteinase 2 (MM

Variable n

MMP-2–positive ex

n (%) �2

Age�60 yrs 37 15 (40.5%) 1.947�60 yrs 37 21 (56.8%)

GenderMale 56 27 (48.2%) 0.017Female 18 9 (50.0%)

DifferentiationWell 52 25 (48.1%) 0.023Mod./poorly 22 11 (50.0%)

Tumor sizeT1/T2 44 20 (45.5%) 0.443T3/T4 30 16 (53.3%)

LN metastasisPositive 26 16 (61.5%) 2.666Negative 48 20 (41.7%)

Clinical stageI/II 36 14 (38.9%) 2.673III/IV 38 22 (57.9%)

RecurrenceYes 25 15 (60.0%) 1.947No 49 21 (42.9%)

Abbreviations as in Table I.

rectly regulate proliferation of some ovarian and

breast cancer cell lines in culture.23,24 However, noreports have examined the relationship betweenCXCR-4 expression and the expression of MMP-2,MMP-9, and Ki-67 using tissue samples from OSCCpatients.

Ki-67 expression with clinicopathologic featurespression Ki-67 LI

P value Mean � SD P value

.812 22.24 � 7.72 .17724.75 � 8.08

.559 23.96 � 8.50 .38322.06 � 5.88

.844 23.09 � 8.17 .50424.45 � 7.50

.068 22.97 � 6.93 .49724.26 � 9.32

.037 23.63 � 6.96 .84023.24 � 9.66

.064 23.71 � 6.73 .82223.29 � 9.04

.366 22.48 � 6.76 .43724.01 � 8.51

and MMP-9 expression with clinicopathologic featuresMMP-9–positive expression

P value n (%) �2 P value

.163 18 (48.5%) 0.871 .35122 (59.4%)

.895 27 (48.2%) 3.161 .07513 (72.2%)

.880 30 (57.7%) 0.932 .33410 (45.4%)

.506 25 (56.8%) 0.334 .56315 (50.0%)

.103 19 (73.1%) 5.840 .01621 (43.8%)

.102 18 (50.0%) 0.464 .49622 (57.9%)

.163 15 (60.0%) 0.537 .46325 (51.0%)

4) anditive ex

�2

0.057

0.342

0.039

3.320

4.366

3.438

0.819

P-2)pression

Therefore, we investigated the relationship between

Scale b

OOOOE680 Lee et al. May 2009

CXCR-4 expression, clinicopathologic factors, and ex-pression of MMP-2, MMP-9, and Ki-67. We furtherassessed the potential value of CXCR-4 as a prognosticindicator of survival in patients with OSCC.

MATERIALS AND METHODSPatients and tumor samples

Samples from 74 patients (56 men and 18 women) withOSCC were examined by immunohistochemistry. All tu-mors were surgically removed at the Department of Oraland Maxillofacial Surgery, Seoul National UniversityDental Hospital, South Korea, between 1995 and 2002.The age of the patients ranged from 27 to 93 years, witha mean of 59.2 years. Clinical data obtained from patient

Fig. 1. Representative immunohistochemical staining in orchemokine receptor 4; b, positive expression of matrix metalland d, high expression of Ki-67 from the same tumor area.

Table III. Correlation of CXCR-4 expression withMMP-2, MMP-9, and Ki-67 expression

Variable n

CXCR-4 expression

Positive Negative �2P

value

MMP-2 expressionPositive 36 25 11 2.193 .139Negative 38 20 18

MMP-9 expressionPositive 40 29 11 4.992 .025Negative 34 16 18

Ki-67 LI, mean �SD

25.83 � 7.62 19.88 � 7.16 .001

Abbreviations as in Table I.

charts included age, gender, TNM stage, and recurrence.

Tumors were staged according to the current TNM clas-sification as recommended by the American Joint Com-mittee on Cancer.25 Tumors were re-reviewed by 2 pa-thologists to determine the histologic grade (welldifferentiated, moderately differentiated, or poorly differ-entiated).26 Survival was calculated from the date of di-agnosis until the date of death or last follow-up.

ImmunohistochemistryImmunohistochemical staining was performed using

the streptavidin-biotin-peroxidase complex method. Sec-tions (4 �m thick) were cut from each paraffin wax block,dewaxed, and incubated for 15 minutes in a methanolsolution containing 3% H2O2 to block endogenous perox-idase activity. The slides were placed in 0.01 mol/L citratebuffer (pH 6.0) and heated in a 500-W microwave ovenfor 2 5-minute periods. After washing with phosphatebuffered saline (PBS), sections were incubated in 10%goat nonimmune serum for 30 minutes to reduce nonspe-cific antibody binding. The primary antibodies used were:mouse monoclonal antihuman CXCR-4 (clone 44716;R&D Systems, Minneapolis,MN) at a concentration of 10�g/mL; antihuman Ki-67 (clone MIB-1; Dako, Carpinte-ria, CA) at a concentration of 1 �g/mL; antihumanMMP-2 (clone 8B4; Santa Cruz Biotechnology, SantaCruz, CA) at a concentration of 2 �g/mL; and antihumanMMP-9 (clone 2C3; Santa Cruz Biotechnology) at a con-centration of 2 �g/mL. Sections were incubated with the

amous cell carcinoma, showing: a, positive expression ofinase 9; c, negative expression of matrix metalloproteinase 2;ar � 100 �m.

al squoprote

primary antibodies for 1 hour at room temperature. After

OOOOEVolume 107, Number 5 Lee et al. 681

being washed with PBS, sections were incubated withbiotinylated goat antimouse IgG for 30 minutes. Theywere then washed 3 times with PBS, treated with strepta-vidin-peroxidase reagent for 30 minutes, and rewashedwith PBS 3 times. The reactions were visualized withdiaminobenzidine (Dako) as the chromogen, and sectionswere counterstained with Mayer hematoxylin. Sectionsfrom normal lymph nodes for CXCR-4 and from anHT-1080 (obtained from American Tissue Cell Collec-tion) cell block for MMP-2 and MMP-9 served as positivecontrol samples. Negative control samples were per-formed by replacing the specific primary antibody withmouse IgG isotype (Sigma, St. Louis, MO).

Evaluation of immunohistochemical resultsThe expression of CXCR-4 was categorized in 4

grades according to staining intensities comparing withthat of interstitial infiltrates27,28: score 3 (strong): stain-ing intensity more than interstitial infiltrates; score 2(moderate): staining intensity equal to interstitial infil-trates; score 1 (mild): staining intensity less than inter-stitial infiltrates; and score 0 (negative): no stainingintensity. We also categorized CXCR-4 expression ac-cording to CXCR-4 expression scores: score 1 to 3:CXCR-4 positive; score 0: CXCR-4 negative. ForMMP-2 and MMP-9, expression was defined as posi-tive when �10% of the tumor cells showed cytoplas-mic staining and the staining intensity was more intensethan that in the adjacent normal mucosa.29 For Ki-67,nuclei from over 1,000 tumor cells were counted in 5-8random fields chosen from each sample and the countwas expressed as the percentage of Ki-67–positive cells(Ki-67 labeling index [LI]). The slides were assessedby consensus between 2 investigators with no priorknowledge of the corresponding clinicopathologic data.

Statistical analysisRelationships between CXCR-4, MMP-2, and

MMP-9 and the various clinicopathologic factors wereexamined using the �2 test. The mean value and SD ofKi-67 LIs were calculated and compared with clinico-pathologic factors using the Student t test.

Survival curves were calculated using the Kaplan-Meier method and analyzed using the log rank test. Theexpression of Ki-67 was classified as high at �23.5%and as low at �23.5% for the Kaplan-Meier method.Deaths caused by OSCC were considered as outcomes,and all deaths from other causes were censored. Theroles of each identified prognostic factor (by univariateanalysis) and combined effects (by multivariate analy-sis) were explored using the Cox proportional hazardssurvival analysis model. In the univariate analysis, thehazards ratio and its 95% confidence interval were

calculated for each variable. Forward stepwise proce-

dures were used to identify the combination of influ-encing components that was important in the predictionof prognosis. All statistical analyses were considered tobe significant when the P value was �.05. Statisticalanalyses were performed using SPSS version 12.0.

RESULTSCorrelation between CXCR-4, MMP-2, MMP-9,and Ki-67 expression and clinicopathologicfactors

Tables I and II show the association of several clin-icopathologic factors with CXCR-4, MMP-2, MMP-9,and Ki-67 expression. CXCR-4 expression was positivein 45 cases (Fig. 1, a). CXCR-4 staining significantlycorrelated with lymph node metastasis (P � .037).Patients with old age, poor differentiation, large tumorsize, high clinical stage, and recurrence tended to showhigher CXCR-4 expression, but these correlations didnot reach significance.

MMP-9 was expressed in 40 cases (Fig. 1, b) andcorrelated with lymph node metastasis (P � .016).MMP-2 was expressed in 36 cases, and the mean Ki-67LI was 23.5 (Fig. 1, c and d). There was no significantcorrelation of MMP-2 or Ki-67 expression with clini-copathologic factors.

Correlation between CXCR-4 expression and theexpression of MMP-2, MMP-9, and Ki-67

CXCR-4 expression was significantly higher in caseswith high expression of MMP-9 than in those with lowexpression (Table III; P � .025). Also, high CXCR-4expression was significantly associated with high Ki-67LI (P � .001). Cases with high MMP-2 expression hadhigher CXCR-4 expression, but this relationship wasnot statistically significant (P � .139).

Table IV. Univariate analysis of survival in 74 patientswith oral squamous cell carcinoma, according to theCox proportional hazards model

VariablesHazard

ratio 95% CIP

value

Age (�60/�60 yrs) 1.493 0.725 to 3.075 .277Gender (F/M) 1.383 0.565 to 3.384 .478Differentiation (well/mod.-

poorly)1.430 0.680 to 3.010 .346

Tumor size (T1-T2/T3-T4) 2.851 1.367 to 5.949 .005LN metastasis (negative/positive) 2.437 1.186 to 5.006 .015Clinical stage (I-II/III-IV) 2.895 1.318 to 6.356 .008Recurrence (no/yes) 2.773 1.347 to 5.710 .006CXCR-4 expression (�/�) 4.545 1.730 to 11.905 .002MMP-2 expression (�/�) 1.736 0.836 to 3.607 .139MMP-9 expression (�/�) 2.390 1.093 to 5.227 .029Ki-67 LI (�23.5/�23.5) 1.874 0.891 to 3.941 .091

CI, Confidence interval; other abbreviations as in Table I.

sis; d,

OOOOE682 Lee et al. May 2009

Survival analysisUnivariate Cox proportional hazards survival analy-

sis of 74 OSCCs revealed that a number of factorsindicated worse prognosis (Table IV), including largetumor size (P � .005), positive lymph node metastasis(P � .015), high clinical stage (P � .008), recurrentdisease (P � .006), high CXCR-4 expression (P �

Fig. 2. Survival curves of patients with oral squamous cell carmetalloproteinase 9 expression; c, lymph node (LN) metasta

.002), and high MMP-9 expression (P � .029). Sur-

vival curves calculated by the Kaplan-Meier methodand analyzed using the log rank test (Fig. 2) indicatedthat the survival of patients with high CXCR-4 expres-sion was significantly worse than those with lowCXCR-4 expression (P � .0007). Also, high MMP-9expression (P � .023), large tumor size (P � .003),high clinical stage (P � .005), and positive lymph node

a according to: a, chemokine receptor 4 expression; b, matrixtumor size; and e, clinical stage (Kaplan-Meier method).

cinom

metastasis (P � .011) were significantly associated

OOOOEVolume 107, Number 5 Lee et al. 683

with poorer survival. Multivariate Cox proportionalhazards survival analysis revealed that CXCR-4 expres-sion, recurrence, and tumor size were independentprognostic factors for survival (Table V).

DISCUSSIONAn accurate assessment of the factors influencing

tumor progression and patient prognosis is crucial forestablishing appropriate therapeutic modalities forOSCC. Several molecular markers used for OSCC clas-sification have been investigated, but there have beenfew with great value in predicting patient prognosis. Inthe present study, we investigated the association be-tween CXCR-4, Ki-67, MMP-2, and MMP-9 expres-sion and several clinicopathologic factors in 74 OSCCsamples. First, CXCR-4 expression was found to asso-ciate significantly with lymph node metastasis, MMP-9expression, and Ki-67 expression. Second, we foundthat high CXCR-4 expression, high MMP-9 expression,large tumor size, high clinical stage, and positive lymphnode metastasis were significantly associated withpoorer survival. Finally, we found that CXCR-4 expres-sion, recurrence, and tumor size were independentprognostic factors for survival by multivariate Coxproportional hazards survival analysis. These data sug-gested that CXCR-4 could be a valuable prognosticmarker for OSCC.

The reported incidence of CXCR-4 expression variesfrom 28.6% to 100% in OSCCs.13,14,28,30,31 We foundCXCR-4 expression in 59.5% of OSCCs. Similarly,Almofti et al.14 found CXCR-4 expression of 57.4% inOSCCs. These differences might result from the use ofthe different evaluation criteria for CXCR-4 positivityand/or the use of different clones of CXCR-4 antibod-ies.

In the present study, CXCR-4 expression correlatedwith lymph node metastasis, in agreement with work byIshikawa et al.13 and Almofti et al.14 However, thelatter showed that CXCR-4 expression correlated withtumor recurrence as well as lymph node metastasis.Katayama et al.,28 in a study using head and necksquamous cell carcinoma tissue, reported that patientswith advanced neck status and patients who developeddistant metastases showed significantly higher CXCR-4expression. Taking all of these results together,CXCR-4 expression clearly correlates with lymph nodemetastasis, but any association with other clinicopath-ologic factors seems to be variable and unclear. Be-cause reports on the relationship between CXCR-4expression and clinicopathologic factors are rare, anadditional larger-scale study would help elucidate anyassociations.

Recently, the association of CXCR-4 activity with

MMPs, such as MMP-2 and MMP-9, has been re-

ported. Brand et al.19 and Chinni et al.20 showed thatCXCR-4 activation up-regulates MMP-2 expression inprostate cancer cells and MMP-9 expression in colo-rectal cancer cells, respectively. Samara et al.,18 in astudy using OSCC cells, reported that CXCR-4 activa-tion by the treatment of CXCL-12 results in an increasein the secretion of MMP-9, not MMP-2, in vitro. How-ever, there have been no reports on such an associationin OSCC tissue samples. The present study does sug-gest a significant correlation between CXCR-4 expres-sion and MMP-9 expression in OSCC. To our knowl-edge, this is the first study to demonstrate theassociation of CXCR-4 expression with MMPs inOSCC tissue samples.

We recently reported that OSCC cells CXCR-4knocked-down by small interfering RNA grew signifi-cantly slower than control cells.32 In agreement withthis, the present study showed a significant correlationbetween CXCR-4 expression and Ki-67 expression inOSCC. These data indicate that CXCR-4 could directlyor indirectly regulate the proliferation of oral cancercells, and CXCR-4 down-regulation might be an effec-tive anticancer therapy in oral cancer. Additional studyon intracellular signaling pathways between these mol-ecules could strengthen the understanding of the bio-logic behaviors of oral cancer.

Katayama et al.,28 in their study using head and necksquamous cell carcinoma tissue, reported that thecause-specific survival of patients with CXCR4 expres-sion was significantly shorter and that CXCR4 positiv-ity was an independent factor for cause-specific deathin multivariate analysis. In agreement with this, ourstudy showed that CXCR-4 expression and tumor sizewere independent prognostic factors for poor survivalin OSCC in univariate and multivariate survival anal-ysis. The multiple functions of CXCR-4, such as en-hancement of proliferation and cell motility, may speedtumor growth and microinvasiveness in OSCC, pro-moting lymph node or distant metastasis of oral cancer.Therefore, high CXCR-4 expression in a patient withOSCC should warrant a more radical mode of treat-ment.

Table V. Multivariate analysis of survival in 74 pa-tients with oral squamous cell carcinoma, according tothe Cox proportional hazards model

VariablesHazard

ratio 95% CIP

value

CXCR-4 expression (�/�) 3.885 1.469 to 10.277 .006Recurrence (no/yes) 2.195 1.051 to 4.582 .036Tumor size (T1-T2/T3-T4) 1.430 0.993 to 4.450 .050

Abbreviations as in Tables I and IV.

OOOOE684 Lee et al. May 2009

In conclusion, the present study revealed thatCXCR-4 expression is a significant prognostic indica-tor for poor outcome in OSCC and may provide a newmethod for guiding OSCC treatment. Furthermore, theinhibition of CXCR-4 is a possible molecular approachto the treatment of OSCC.

REFERENCES1. Sankaranarayanan R, Masuyer E, Swaminathan R. Head and

neck cancer: a global perspective on epidemiology and survival.Anticancer Res 1998;18:4779-86.

2. Carvalho AL, Ikeda MK, Magrin J, Kowalski LP. Trends of oraland oropharyngeal cancer survival over five decades in 3267patients treated in a single institution. Oral Oncol 2004;40:71-6.

3. Carvalho AL, Magrin J, Kowalski LP. Sites of recurrence in oraland oropharyngeal cancers according to the treatment approach.Oral Dis 2003;9:112-8.

4. Shah JP, Lydiatt W. Treatment of cancer of the head and neck.Cancer J Clin 1995;45:352-68.

5. Arya M, Patel HR, Williamson M. Chemokines: key players incancer. Curr Med Res Opin 2003;19:557-64.

6. Muller A, Homey B, Soto H, Ge N, Catron D, Buchanan ME, etal. Involvement of chemokine receptors in breast cancer metas-tasis. Nature 2001;410:50-6.

7. Scotton CJ, Wilson JL, Milliken D, Stamp G, Balkwill FR.Epithelial cancer cell migration: a role for chemokine receptors?Cancer Res 2001;61:4961-5.

8. Taichman RS, Cooper C, Keller ET, Pienta KJ, Taichman NS,McCauley LK. Use of the stromal cell-derived factor-1/CXCR4pathway in prostate cancer metastasis to bone. Cancer Res2002;62:1832-7.

9. Schrader AJ, Lechner O, Templin M, Dittmar KE, Machtens S,Mengel M, et al. CXCR4/CXCL12 expression and signalling inkidney cancer. Br J Cancer 2002;86:1250-6.

10. Zhou Y, Larsen PH, Hao C, Yong VW. CXCR4 is a majorchemokine receptor on glioma cells and mediates their survival.J Biol Chem 2002;277:49481-7.

11. Kijima T, Maulik G, Ma PC, Tibaldi EV, Turner RE, Rollins B,et al. Regulation of cellular proliferation, cytoskeletal function,and signal transduction through CXCR4 and c-Kit in small celllung cancer cells. Cancer Res 2002;62:6304-11.

12. Hwang JH, Hwang JH, Chung HK, Kim DW, Hwang ES, SuhJM, et al. CXC chemokine receptor 4 expression and function inhuman anaplastic thyroid cancer cells. J Clin Endocrinol Metab2003;88:408-16.

13. Ishikawa T, Nakashiro K, Hara S, Klosek SK, Li C, Shintani S,et al. CXCR4 expression is associated with lymph-node metas-tasis of oral squamous cell carcinoma. Int J Oncol 2006;28:61-6.

14. Almofti A, Uchida D, Begum NM, Tomizuka Y, Iga H, YoshidaH, et al. The clinicopathological significance of the expression ofCXCR4 protein in oral squamous cell carcinoma. Int J Oncol2004;25:65-71.

15. Birkedal-Hansen H, Moore WGI, Bodden MK, Windsor LJ,Birkedal-Hansen B, DeCarlo A, et al. Matrix metalloproteinases.A review. Crit Rev Oral Biol Med 1993;4:197-250.

16. Gao ZB, Duan YQ, Zhang L, Chen DW, Ding PT. Expression ofmatrix metalloproteinase 2 and its tissue inhibitor in oral squa-mous cell carcinoma. Int J Mol Med 2005;16:599-603.

17. Hong SD, Hong SP, Lee JI, Lim CY. Expression of matrixmetalloproteinase-2 and -9 in oral squamous cell carcinomaswith regard to the metastatic potential. Oral Oncol 2000;36:207-13.

18. Samara GJ, Lawrence DM, Chiarelli CJ, Valentino MD, Lyub-

sky S, Zucker S, et al. CXCR4-mediated adhesion and MMP-9secretion in head and neck squamous cell carcinoma. Cancer Lett200;214:231-41.

19. Brand S, Dambacher J, Beigel F, Olszak T, Diebold J, Otte JM,et al. CXCR4 and CXCL12 are inversely expressed in colorectalcancer cells and modulate cancer cell migration, invasion andMMP-9 activation. Exp Cell Res 2005;310:117-30.

20. Chinni SR, Sivalogan S, Dong Z, Filho JC, Deng X, BonfilRD, et al. CXCL12/CXCR4 signaling activates Akt-1 andMMP-9 expression in prostate cancer cells: the role of bonemicroenvironment–associated CXCL12. Prostate. 2006;66:32-48.

21. Singh S, Singh UP, Grizzle WE, Lillard JW Jr. CXCL12-CXCR4interactions modulate prostate cancer cell migration, metallopro-teinase expression and invasion. Lab Invest 2004;84:1666-76.

22. Schluter C, Duchrow M, Wohlenberg C, Becker MH, Key G,Flad HD, et al. The cell proliferation-associated antigen of anti-body Ki-67: a very large, ubiquitous nuclear protein with numer-ous repeated elements, representing a new kind of cell cycle-maintaining proteins. J Cell Biol 1993;123:513-22.

23. Hall JM, Korach KS. Stromal cell-derived factor 1, a novel targetof estrogen receptor action, mediates the mitogenic effects ofestradiol in ovarian and breast cancer cells. Mol Endocrinol2003;17:792-803.

24. Lapteva N, Yang AG, Sanders DE, Strube RW, Chen SY.CXCR4 knockdown by small interfering RNA abrogates breasttumor growth in vivo. Cancer Gene Ther 2005;12:84-9.

25. Greene FL, Page DL, Fleming ID, Fritz A, Balch CM, HallerDG, Morrow M. AJCC cancer staging manual. 6th ed. Chicago:Springer; 2002.

26. Neville BW, Damm DD, Allen CM, Bouquot JE. Oral andmaxillofacial pathology. 2nd ed. Philadelphia: Saunders; 2002.

27. Kato M, Kitayama J, Kazama S, Nagawa H. Expression patternof CXC chemokine receptor-4 is correlated with lymph nodemetastasis in human invasive ductal carcinoma. Breast CancerRes 2003;5:R144-50.

28. Katayama A, Ogino T, Bandoh N, Nonaka S, Harabuchi Y.Expression of CXCR4 and its down-regulation by IFN-gamma inhead and neck squamous cell carcinoma. Clin Cancer Res2005;11:2937-46.

29. Juuti A, Lundin J, Nordling S, Louhimo J, Haglund C. EpithelialMMP-2 expression correlates with worse prognosis in pancreaticcancer. Oncology. 2006;71:61-8.

30. Muller A, Sonkoly E, Eulert C, Gerber PA, Kubitza R, SchirlauK, et al. Chemokine receptors in head and neck cancer: associ-ation with metastatic spread and regulation during chemother-apy. Int J Cancer 2006;118:2147-57.

31. Delilbasi CB, Okura M, Iida S, Kogo M. Investigation ofCXCR4 in squamous cell carcinoma of the tongue. Oral Oncol2004;40:154-7.

32. Hong JS, Pai HK, Hong KO, Kim MA, Kim JH, Lee JI, et al.CXCR-4 knockdown by small interfering RNA inhibits cellproliferation and invasion of oral squamous cell carcinoma cells.J Oral Pathol Med 2009;38:214-9.

Reprint requests:

Seong-Doo HongDepartment of Oral Pathology,School of DentistrySeoul National University28 Yeonkeun-dong, Chongro-guSeoul 110-749Korea

hongsd@snu.ac.kr.