Journal of Surgical Research 171, 143–150 (2011)doi:10.1016/j.jss.2010.03.001

The CXCR4-CXCL12 Pathway Facilitates the Progression of Pancreatic

Cancer Via Induction of Angiogenesis and Lymphangiogenesis

Kai Cui, M.M.,*,3 Wenhua Zhao, M.M.,†,3 Changliang Wang, M.M.,* Ailiang Wang, M.M.,‡ Bo Zhang, M.D.,*Wuyuan Zhou, M.M.,* Jinming Yu, M.D.,* Ziqiang Sun, M.M.,‡,2 and Sheng Li, M.D.*,1

*Shandong Tumor Hospital, Ji’nan, China; †Shandong Qianfoshan Hospital, Ji’nan, China; and ‡Affiliated Hospital of Ji’ning MedicalCollege, Ji’ning, China

Submitted for publication November 6, 2009

Background. This study reports the influence ofCXCL12 and its receptor CXCR4 on the progression ofpancreatic cancer and illuminates the correlation be-tween the CXCL12/CXCR4 axis and the angiogenesisand lymphangiogenesis of pancreatic adenocarcinoma(PAC).

Methods. A total of 30 patients with pancreatic can-cer participated in the current study. The expressionof CXCL12 and CXCR4 in cancerous tissues, paracan-cerous tissues, normal pancreas, and lymph nodessurrounding the pancreas were investigated usingreal-time PCR and immunohistochemistry, respec-tively. In addition, we assessed microvessel density(MVD) and microlymphatic vessel density (MLVD) intumor tissues using immunohistochemistry.

Results. CXCL12 expression in tumor tissues wassignificantly lower than that of paracancerous tissues,normal pancreas, and lymph nodes. In contrast,CXCR4 expression in cancerous tissues was consider-ably higher than that of normal pancreas. Addition-ally, a significant correlation between the expressionpattern of the CXCL12/CXCR4 axis and clinicopatho-logic features, such as lymph node metastasis, wasidentified. Furthermore, we found that CXCL12expression was significantly associated with MVD butnot significantly associated with MLVD, while CXCR4expression was significantly associated with MLVDbut not significantly associated with MVD.

1 To whom correspondence and reprint requests should beaddressed Shandong Tumor Hospital, Jinan 250117, ShandongProvince, China. E-mail: drlisheng@sohu.com.

2 To whom correspondence and reprint requests should beaddressed at Affiliated Hospital of Jining Medical College, Ji’ning272111, China. E-mail: sunziqiang1893@sina.com.

3 These authors contributed equally to this work

143

Conclusions. The chemotactic interaction betweenCXCR4 and its ligand CXCL12 may be a critical eventduring the progression of pancreatic cancer. Theunderlying mechanism may be the induction of angio-genesis and lymphangiogenesis regulated by the inter-action of CXCL12 and CXCR4. � 2011 Elsevier Inc. All rights

reserved.

Key Words: chemokines; chemokine receptors; angio-genesis; lymphangiogenesis; pancreatic neoplasms.

INTRODUCTION

Pancreatic cancer is the fourth leading cause of can-cer deaths in the United States, causing 35,240 deathsbased on U.S. Mortality Data, 1969–2006 [1]. In con-trast to gastrointestinal malignancies, the overall 5-ysurvival after a diagnosis of pancreatic adenocarcinomais abysmal [2]. This exceptionally poor prognosis islargely attributed to the fact that, generally, pancreaticcancer has already metastasized by the time it is diag-nosed. Although some of the key genetic mutationspresent in pancreatic cancer have been identified, ourunderstanding of the signaling events that mediatethe progression from pancreatic cancer precursors toinvasive, metastatic tumors is extremely limited [3].

The predilection of metastases for specific organsmay depend on a variety of factors [4, 5]. According toone theory, migrating tumor cells can enter any tissuebut do not form a metastasis unless requirements forgrowth are satisfied [6]. A second theory suggests thattissue-specific adhesion molecules on endothelial cellsselect migrating cells that can attach and form a preme-tastatic nucleus of cells [7]. The most recent theory pro-poses that chemoattractants produced by stromal or

0022-4804/$36.00� 2011 Elsevier Inc. All rights reserved.

JOURNAL OF SURGICAL RESEARCH: VOL. 171, NO. 1, NOVEMBER 2011144

immune cells lead invasive tumor cells to tissues for po-tential secondary growth [8].

Chemokines, a group of homologous yet functionallydivergent proteins, directly mediate the migrationand activation of leukocytes and play a role in regulat-ing angiogenesis [9]. Chemokines also maintain im-mune homeostasis and secondary lymphoid organarchitecture [10]. Several chemokines are known tohave antitumor activity. Tumor rejection has beennoted in various murine tumor models in which tumorcells have been modified with chemokines, includingMIP1, RANTES, lymphotactin, TCA3, JE/MCP-1/MCAF, MIP3, MIP3ß, and IP-10 [11–19]. Among thechemokines, the role of CXCL12 and CXCR4 in theprogression of various cancers has been intensivelystudied. CXCR4 is highly expressed in human breastcancer cells, malignant breast tumors, andmetastases. CXCL12, the ligand binding to CXCR4,induces chemotactic and invasive responses, includingactin polymerization and pseudopodia formation [20].CXCR4 is expressed in several other tumor types andis correlated with lymph node metastases and poorprognosis in esophageal cancer [21], non-small-celllung cancer [22], and squamous cell carcinoma of thehead and neck [23, 24]. Although the importance ofCXCR4 has been identified in these cancers, fewreports describe its function in pancreatic cancer.Based on previous findings, we hypothesized that theCXCL12/CXCR4 axis plays an important role in theprogression and metastasis of pancreatic carcinoma.

In this study, we evaluated the expression of theCXCL12/CXCR4 axis and its relationship with tumorstage and grade in 30 patients with pancreatic cancer.We also analyzed the relationship of CXCL12 andCXCR4 expression with intratumoral and peritumoralMVD and MLVD in pancreatic cancer. Our data sug-gest that the CXCL12/CXCR4 axis mediates potent re-sponses during the progression and metastasis ofpancreatic cancer.

MATERIAL AND METHODS

Tissue Samples

Tissue samples were obtained from 30 patients who underwentmacroscopically curative resection at Shandong Tumor Hospital be-tween 2005 and 2007. Samples of the pancreatic tumor, paracancer-ous tissues, normal pancreas, and lymph nodes surrounding thepancreas were immediately frozen in liquid nitrogen or formalin-fixed after surgery and then embedded in paraffin. Sections fromeach case were stained with hematoxylin and eosin (H and E) for his-tologic examination according to the tumor-node-metastasis (TNM)classification system. Tissue was collected based on the protocol ap-proved by the Ethics Committee of the Medical Faculty of ShandongTumor Hospital. All patients had complete clinical and pathologicdata. The patients were comprised of 17 men and 13 women with a me-dian age of 57.2 y (range 35–78 y). No patients received preoperativechemotherapy or radiotherapy. Among the 30 patients, 17 well-

differentiated and 13 poorly-differentiated cases were identified.The study included 12 Union Internationale Contra Cancrum(UICC) stages I to II patients and 18 patients who were stages IIIto IV.

Immunohistochemical Staining

Tissue sections were analyzed using streptavidin-peroxidase im-munohistochemical staining. We used 4-mm-thick sections of repre-sentative blocks with antibodies against the chemokines: CXCL12(dilution 1:200), CXCR4 (dilution 1:200), VEGFR-3 (dilution 1:250),and CD34 (dilution 1:80) (DAKO Cytomation, Glostrup, Denmark)The sections were deparaffinized and rehydrated. After blocking ofendogenous peroxidase with methanol containing 0.3% H2O2, the sec-tions were autoclaved at 121 �C for 10 min in citrate buffer (10 mmol/Lsodium citrate, pH¼ 6) for antigen retrieval. After blocking with nor-mal goat serum, the sections were reacted overnight with primary an-tibodies, which were appropriately diluted. The sections were thenreacted sequentially with biotin-conjugated anti-mouse immunoglob-ulin G antibodies (Vector Laboratories, Burlingame, CA) and Vectas-tain Elite ABC reagent (Vector Laboratories). Diaminobenzidine wasused as the chromogen, and the nuclei were counterstained with he-matoxylin. One breast cancer sample was used as a positive control,and phosphate buffered solution (PBS) was the negative control.Each section was analyzed by two double-blinded pathologists interms of the staining intensity and the proportion of positive tumorcells. The upper quartile was defined as the cutoff point.

Microvessels were detected by morphological observation and im-munohistochemical labeling using the endothelial marker CD34 forMVD evaluation. All independent CD34-positive vessels werecounted regardless of the presence of an identifiable lumen. We em-ployed the following methods to assess MVD. First, the area withthe most intense vascularization was determined under low magnifi-cation. Then, average MVD was analyzed by randomly selecting fivefields per tumor at 4003 magnification. Blood vessels with lumenscontaining more than eight red cells or with muscular layers were ex-cluded. For each case, the number of CD34-positive vessel structuresin five high power fields was recorded, and the average value was theMVD. The immunostaining results were assessed by two pathologistswho were blind to the clinicopathologic findings. Meanwhile, VEGFR-3 was used for MLVD evaluation. The process used to calculate MLVDwas the same as that of the MVD assay mentioned above.

RT-PCR Analysis

Real-time PCR was carried out according to the manufacturer’s in-structions (Takara, Dalian, China). In brief, RNA extraction from fro-zen human specimens was performed using the acid guanidiniumthiocyanate method. RNA was dissolved in diethylpyrocarbonate(DEPC)-treated water (0.1% DEPC was added to water overnightand then autoclaved for 20 min to destroy DEPC). The preparedRNA (1 mg) was mixed with reverse transcription reagents at a totalvolume of 20 mL and incubated for 30 min at 42 �C to produce first-strand cDNA. A total of 1 mL cDNA was used for PCR amplification(Invitrogen, Carlsbad, CA). The primer sequences were as follows:

CXCR4 forward: 50-AGCTGTTGGTGAAAAGGTGGTCTATG-30,CXCR4 reverse: 50-GCGCTTCTGGTGGCCCTTGGAGTGTG-30;CXCL12 forward: 50-CCGCGCTCTGCCTCAGCGACGGGAAG-30,CXCL12 reverse: 50-CTTGTTTAAAGCTTTCTCCAGGTACT-30;b-actin forward: 50-GTGGGGCGCCCCAGGCACCA-30;b-actin reverse: 50-CTCCTTAATGTCACGCACGATTT-30.

Following the manufacturer’s instructions, reverse transcriptionwas performed at 42 �C in the presence of 5 units AMV reverse tran-scriptase and 1 mg RNA for 60 min. AMV RT inactivation and RNA/cDNA/primer denaturation were performed at 95 �C for 5 min to acti-vate modified Taq polymerase followed by 40 cycles at 95 �C for 10min, 95 �C for 10 s, 56 �C for 1 min, and 1 cycle at 72 �C for 35 s.

CUI ET AL.: CXCR4-CXCL12 PATHWAY AND PANCREATIC CANCER 145

TABLE 1

CXCL12 and CXCR4 Expression by Immunohistochemical Staining

CXCL12 expression CXCR4 expression

Group Negative Positive n (%) P Negative Positive n (%) P

Cancerous tissues 26 4 (13.3) 6 24 (80.0)Paracancerous tissues 16 14 (46.7) 0.011 9 21 (70.0) 0.6Normal pancreas 13 17 (56.7) 0.001 22 8 (26.7) <0.001Lymph nodes 15 15 (50.0) 0.006 8 22 (73.3) 0.8

JOURNAL OF SURGICAL RESEARCH: VOL. 171, NO. 1, NOVEMBER 2011146

The PCR product was separated by 2% agarose gel electrophoresis.The gels were viewed using UV transillumination and photographedby a Kodak 120 gel imaging system.

Statistical Analysis

The data were analyzed using the SPSS software program (version13.0 for Windows; SPSS Inc., Chicago, IL). Comparison among differ-ent groups was performed by c2 test, Fisher’s exact probabilities, one-way ANOVA with Bonferroni t-test and Spearman rank correlationcoefficient. P < 0.05 was considered statistically significant. Afterthe step of determining statistical power of 0.8, we calculated the min-imal required sample size is thirty cases in our study.

RESULTS

CXCL12 and CXCR4 Expression in Pancreatic Cancer

The levels of CXCL12 and CXCR4 mRNA in the pri-mary tumor, paracancerous tissues, normal pancreas,and lymph nodes surrounding the pancreas were as-sessed using RT-PCR. Immunohistochemical stainingwas performed to further determine the levels and loca-tions of CXCL12 and CXCR4 protein (Fig. 1). The ex-pression of CXCL12 was lower in the pancreatictumor tissues compared with those of other tissues(Table 1). As Table 1 shows, 80.0% of cancerous tissues,70.0% of paracancerous tissues, and 73.3% of lymphnode tissues adjacent to the pancreas had positiveexpression of CXCR4. These results were significantlydifferent from that of normal pancreas. Positive expres-sion of CXCR4 was also detected in the vascular smoothmuscle of the pancreas (Fig. 1I), fat surrounding theperipancreatic lymph nodes (Fig. 1J), and an ‘‘island’’of pancreatic cancer (Fig. 1K).

The level of CXCL12 mRNA was low in tumor tissuesand moderate in normal pancreas (Fig. 2), while weakexpression of CXCL12 was found in pancreatic cancer(Fig. 1B). A significant difference was found betweenthe pancreatic cancer and normal pancreas (P < 0.01)(Tables 1 and 2). Moderate expression of CXCL12 wasobserved in the paracancerous tissues adjacent to the

FIG. 1. Immunohistochemical staining of CXCL12 in normal pancrnode (D). Original magnification: 3100. Immunohistochemical staining otissues (G), lymph node (H), blood vessels of cancerous tissues (I), fat surtissues (K). Original magnification: 3100.

pancreas at the mRNA level, and this wassignificantly different from that of the pancreaticcancer group (P < 0.05) (Table 2). CXCL12 mRNA wasdetected at significantly high levels in paracanceroustissues, normal pancreas and lymph nodes (Table 2).The levels of CXCL12 protein exhibited the same trendas those of the mRNA (Fig. 1A<194>D). CXCR4 expres-sion tended to be opposite that of CXCL12 at both themRNA and protein levels (Tables 1 and 2). Comparedwith normal pancreas, the expression of CXCR4 incancerous tissues, paracancerous tissues, and lymphnodes was significantly higher (P < 0.001) (Table 2,Fig. 2).

CXCL12/CXCR4 Interaction and Clinicopathologic Factors of

Pancreatic Cancer

To define the role of CXCL12 and CXCR4 in the pro-gression of pancreatic cancer, we analyzed the associa-tion of CXCL12 and CXCR4 expression profiles withtumor grade and stage. This analysis revealed thathigher CXCR4 expression and the cytoplasmic localiza-tion of CXCR4 were significantly associated with thelymph node status of the tumor (Table 3). In addition,CXCR4 expression also correlated with stage in pancre-atic cancer patients (P ¼ 0.05) (Table 3). No definitivecorrelation was observed between CXCL12 expressionand progression of pancreatic cancer (Table 3).

Lymphatic and Blood Vessels in Pancreatic Cancer

Table 4 shows the varying degrees of angiogenesisand lymphangiogenesis, and the relationships betweenthese variables and clinical outcome were determined.Compared with patients with lower intratumoralMVD, patients with higher intratumoral MVD ex-hibited a significantly lower degree of differentiationbut a higher class of TNM staging (P < 0.05) (Table 4).However, no relationship was found between lymphnode metastasis and blood vessel formation (P > 0.05)

eas (A), cancerous tissues (B), paracancerous tissues (C), and lymphf CXCR4 in normal pancreas (E), cancerous tissues (F), paracancerousrounding peripancreatic lymph nodes (J), and an ‘‘island’’ of cancerous

FIG. 2. (A) CXCL12 mRNA expression; (B) CXCR4 mRNA expression.

CUI ET AL.: CXCR4-CXCL12 PATHWAY AND PANCREATIC CANCER 147

(Table 4). Additionally, it was found that MLVD statuscorrelated with tumor stage and grade. The formationof lymphatic vessels significantly increased in patientswith higher TNM staging and lymph node metastasis,but no significant change was observed in patientswith a lower degree of differentiation (Figs. 3 and 4).

Analysis of the Association of CXCL12/CXCR4 with MVD/

MLVD

As shown in Table 5, CXCL12 protein was negative in26 samples, and the MVD in these negative sampleswas higher than that in positive samples (P ¼ 0.022).No significant correlation between the expression ofCXCL12 protein and MLVD of pancreatic cancer wasfound (P > 0.05) (Table 5). CXCR4 protein was positivein 24 samples, and the MLVD in these positive sampleswas higher than that in the negative group (P ¼ 0.003)(Table 5). These results indicate that lower expressionof CXCL12 protein was strongly associated with angio-genesis of pancreatic cancer, while higher expression ofCXCR4 protein was significantly associated with lym-phangiogenesis of pancreatic cancer.

TABLE 3

Correlations Between Expression of CXCL12 and

DISCUSSION

Chemokine molecules constitute a superfamily of in-ducible secreted proinflammatory proteins [25–28] thatare involved in a variety of immune responses. These

TABLE 2

CXCL12 and CXCR4 mRNA Expression by RT-PCR

CXCL12 expression CXCR4 expression

GroupCXCL12/b-actin P

CXCR4/b-actin P

Cancerous tissues 0.263 6 0.254 0.789 6 0.300Paracancerous

tissues0.429 6 0.136 0.003 0.701 6 0.291 0.3

Normal pancreas 0.437 6 0.098 <0.001 0.236 6 0.199* <0.001Lymph nodes 0.425 6 0.187 0.006 0.700 6 0.322 0.3

*P < 0.001 compared with paracancerous tissues and lymph nodesvalue by one-way ANOVA and Bonferroni t-test.

molecules primarily act as chemoattractants andactivators of specific types of leukocytes [25, 29, 30],and they mediate these functions by binding to G-protein-coupled receptors [31]. It is becoming increas-ingly evident that chemokines play an integral role ininitiating specific immune responses [32]. One suchchemokine (CXCL12) is found on the high endothelialvenules and within T-cell zones of the spleen and lymphnodes [17, 33–35], and it is involved in the recruitmentof naive T cells and dendritic cells (DCs). In the lymphnodes, CXCL12 plays an important role in theinitiation of immune responses by co-localizing naiveT cells with DC-presented antigens [12, 15, 36, 37].The receptor for this ligand, CXCR4, is expressed onall naive T cells, some memory T cells, B cells, andmature dendritic cells and plays a central role inlymphocyte trafficking and homing to lymph nodes[21, 38]. Recent studies have shown the involvementof the CXCL12/CXCR4 axis in the progression ofseveral types of cancer [19, 39, 40]. For example,Hassan [41] et al. reported high levels of CXCL12/CXCR4 expression in breast cancer cells and linked re-ceptor expression to the metastatic destination of tumorcells. However, correlations between the CXCL12/

CXCR4 and Clinicopathological Factors of PancreaticCancer

Factors n

CXCR4expressionPositive (%) P

CXCL12expressionPositive (%) P

Histology 0.3 0.8Well-differentiated 17 12 (70.6) 3 (17.6)Moderately-/poorly-differentiated

13 12 (92.3) 1 (7.8)

TNM stage 0.05 0.3I-II 12 7 (58.3) 3 (33.3)III-IV 18 17 (94.4) 1 (5.6)

Lymph node metastasis 0.004 0.9Negative 12 6 (50.0) 1 (8.3)Positive 18 18 (100.0) 3 (16.7)

TABLE 4

Correlations between MVD/MLVD and Clinicopatho-logic Factors of Pancreatic Cancer

n MVD P MLVD P

Histology <0.001 0.2Well-differentiated 17 54 6 4.3 5 6 5.9Moderately-/poorly-differentiated

13 65 6 6.5 8 6 6.4

TNM stage <0.001 0.017I-II 12 52 6 7.4 4 6 2.9III-IV 18 63 6 7.3 8 6 4.9

Lymph node metastasis 0.5 0.009Negative 12 52 6 7.4 4 6 2.9Positive 18 54 6 6.9 9 6 5.7

FIG. 3. CD34-positive micro-vascular endothelial cells in pancre-atic cancer. Original magnification: 3100.

JOURNAL OF SURGICAL RESEARCH: VOL. 171, NO. 1, NOVEMBER 2011148

CXCR4 axis and clinical features of pancreatic cancerhave not been extensively studied. Therefore, in ourstudy, we evaluated the expression of CXCL12/CXCR4 and found a critical relationship betweenCXCL12/CXCR4 and tumor stage and grade in pancre-atic cancer.

For patients with pancreatic carcinoma, CXCL12 ex-pression was reduced in tumor tissues, but significantlevels were detected in paracancerous tissues, normalpancreas, and lymph nodes. CXCR4 expression showedthe opposite trend. CXCR4 was expressed in 80.0% ofpancreatic carcinoma samples but only 26.7% of normalsamples. In addition, we found significant differencesbetween positive and negative CXCL12/CXCR4 sam-ples in the following clinicopathological features: (1)lymph node metastasis and (2) tumor TNM staging.

To elucidate the underlying association betweenCXCL12/CXCR4 expression and clinicopathologic fea-tures, we assessed MVD and MLVD in tumor tissues.We found that CXCL12 expression was significantly as-sociated with the formation of blood vessels, whereas ithad no obvious relationship with MLVD. The expres-

FIG. 4. Immunohistochemical staining of VEGF-R3 to identify miccreatic cancer (B). Original magnification: 3100.

sion of CXCR4 was higher in patients with higherMLVD compared to those with lower MLVD. CXCR4had no relationship with MVD.

Angiogenesis and lymphangiogenesis are requiredfor many pathological processes, including tumorgrowth, metastasis, and physiologic organ/tissue main-tenance. In general, the molecular mechanisms thatcontrol carcinoma progression and metastasis are re-lated to mutations of various oncogenes, tumor sup-pressor genes, metastasis suppressor genes, growthfactors, and their receptors, including Src, Ras, p16,KiSS-1, Nm23, FasL, vascular endothelial growth fac-tor (VEGF), basic fibroblast growth factor (bFGF), andinterleukin (IL)-6 [19, 42–44]. These abnormalitiesaffect the downstream signal transduction pathwaysinvolved in the control of cell growth and othermalignant properties, such as tumor staging anddegree of tumor differentiation. Interestingly, one ofthe most recently recognized events in this processinvolves the interaction between chemokines and

ro-lymph vessel endothelial cells in paracancerous tissues (A) and pan-

TABLE 5

Correlations Between CXCL12/CXCR4 and MVD/MLVD

n MVD P MLVD P

CXCL12 0.02 0.5Negative 26 64 6 6.9 7.9 6 5.3Positive 4 55 6 7.0 5.9 6 3.1

CXCR4 0.8 0.003Negative 6 53 6 4.8 6.1 6 5.8Positive 24 54 6 4.1 12.1 6 4.0

CUI ET AL.: CXCR4-CXCL12 PATHWAY AND PANCREATIC CANCER 149

their receptors. Several studies have found that thechemokine receptor was highly expressed in breastand ovarian carcinomas, and the interaction betweenthe receptor and its ligand resulted in chemotaxis, ordirected migration of tumor cells from their primarysite via the circulation to preferential sites ofmetastasis [45–48]. These studies strongly supportour hypothesis. The interaction between CXCR4 andCXCL12 may play crucial roles in the metastasis andprogression of pancreatic cancer by its effects on theformation of new blood vessels and lymphatic vessels.

In conclusion, our data suggest that the chemotacticinteraction between CXCR4 and its ligand CXCL12may be a critical event in the progression of pancreaticcancer. A potential mechanism of action may be the in-duction of angiogenesis and lymphangiogenesis by can-cer cells. This hypothesis is supported by findings thatthe expression pattern of CXCL12 and CXCR4 in pan-creatic cancer tissue was significantly correlated withclinicopathologic features. Our data also suggest thatthe CXCL12/CXCR4 axis is associated with the forma-tion of lymphatic vessels and blood vessels induced bypancreatic cancer cells. Additional work is under wayto determine the pathways responsible for tumor cellchemokine and/or chemokine receptor-associated an-giogenesis and lymphangiogenesis. It is likely that con-trolling such a poor prognostic feature would enablemore successful loco-regional tumor control and im-prove survival in patients with pancreatic cancer.

ACKNOWLEDGMENTS

This work was financially supported by the National Natural Sci-ence Foundation of China (30571712 and 30810403081), and the De-partment of Science and Technology of Shandong Province of China(2007GG20002022). The authors thank the pathologists Wei-xiaZhong, Dian-bin Mu, and Lan-ping Sun for their excellent technicalsupport in evaluating the results of the immunohistochemicalstaining.

REFERENCES

1. Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2009. CACancer J Clin 2009;59:225.

2. Jemal A, Murray T, Samuels A, et al. Cancer statistics, 2003.CA Cancer J Clin 2003;53:5.

3. Sohn TA, Yeo CJ. The molecular genetics of pancreatic ductalcarcinoma: A review. Surg Oncol 2000;9:95.

4. Fidler IJ, Hart IR. Biological diversity in metastatic neoplasms:Origins and implications. Science 1982;217:998.

5. Liotta LA. An attractive force in metastasis. Nature 2001;410:24.

6. Liotta LA, Saidel MG, Kleinerman J. The significance of hema-togenous tumor cell clumps in the metastatic process. CancerRes 1976;36:889.

7. Nicolson GL. Molecular mechanisms of cancer metastasis:Tumor and host properties and the role of oncogenes and sup-pressor genes. Curr Opin Oncol 1991;3:75.

8. Chambers AF, Groom AC, MacDonald IC. Dissemination andgrowth of cancer cells in metastatic sites. Nat Rev Cancer2002;2:563.

9. Baggiolini M, Dewald B, Moser B. Human chemokines: Anupdate. Annu Rev Immunol 1997;15:675.

10. Jung S, Littman DR. Chemokine receptors in lymphoid organhomeostasis. Curr Opin Immunol 1999;11:319.

11. Xu L, Zhou Y, Liu Q, et al. CXCR4/SDF-1 pathway is crucial forTLR9 agonist enhanced metastasis of human lung cancer cell.Biochem Biophys Res Commun 2009;382:571.

12. Bottazzi B, Walter S, Govoni D, et al. Monocyte chemotacticcytokine gene transfer modulates macrophage infiltration,growth, and susceptibility to IL-2 therapy of a murine mela-noma. J Immunol 1992;148:1280.

13. Miwa S, Mizokami A, Keller ET, et al. The bisphosphonateYM529 inhibits osteolytic and osteoblastic changes and CXCR-4-induced invasion in prostate cancer. Cancer Res 2005;65:8818.

14. Sallusto F, Schaerli P, Loetscher P, et al. Rapid and coordinatedswitch in chemokine receptor expression during dendritic cellmaturation. Eur J Immunol 1998;28:2760.

15. Sozzani S, Allavena P, D’Amico G, et al. Differential regulationof chemokine receptors during dendritic cell maturation: Amodel for their trafficking properties. J Immunol 1998;161:1083.

16. Dieu MC, Vanbervliet B, Vicari A, et al. Selective recruitment ofimmature and mature dendritic cells by distinct chemokinesexpressed in different anatomic sites. J Exp Med 1998;188:373.

17. Monterrubio M, Mellado M, Carrera AC, et al. PI3Kgamma acti-vation by CXCL12 regulates tumor cell adhesion and invasion.Biochem Biophys Res Commun 2009;388:199.

18. Ricciardi A, Elia AR, Cappello P, et al. Transcriptome of hypoxicimmature dendritic cells: Modulation of chemokine/receptorexpression. Mol Cancer Res 2008;6:175.

19. Uchida D, Onoue T, Begum NM, et al. Vesnarinone downregu-lates CXCR4 expression via upregulation of Kruppel-like factor2 in oral cancer cells. Mol Cancer 2009;8:62.

20. Muller A, Homey B, Soto H, et al. Involvement of chemokinereceptors in breast cancer metastasis. Nature 2001;410:50.

21. Sasaki K, Natsugoe S, Ishigami S, et al. Expression of CXCL12and its receptor CXCR4 in esophageal squamous cell carcinoma.Oncol Rep 2009;21:65.

22. Reckamp KL, Figlin RA, Burdick MD, et al. CXCR4 expressionon circulating pan-cytokeratin positive cells is associated withsurvival in patients with advanced non-small cell lung cancer.BMC Cancer 2009;9:213.

23. Yoon Y, Liang Z, Zhang X, et al. CXC chemokine receptor-4 an-tagonist blocks both growth of primary tumor and metastasis ofhead and neck cancer in xenograft mouse models. Cancer Res2007;67:7518.

24. Tan CT, Chu CY, Lu YC, et al. CXCL12/CXCR4 promotes laryn-geal and hypopharyngeal squamous cell carcinoma metastasisthrough MMP-13-dependent invasion via the ERK1/2/AP-1pathway. Carcinogenesis 2008;29:1519.

25. Morales J, Homey B, Vicari AP, et al. CTACK, a skin-associatedchemokine that preferentially attracts skin-homing memoryT cells. Proc Natl Acad Sci USA 1999;96:14470.

JOURNAL OF SURGICAL RESEARCH: VOL. 171, NO. 1, NOVEMBER 2011150

26. Zlotnik A, Yoshie O. Chemokines: A new classification systemand their role in immunity. Immunity 2000;12:121.

27. Campbell JJ, Butcher EC. Chemokines in tissue-specific andmicroenvironment-specific lymphocyte homing. Curr OpinImmunol 2000;12:336.

28. Butcher EC, Williams M, Youngman K, et al. Lymphocyte traf-ficking and regional immunity. Adv Immunol 1999;72:209.

29. Klein RS, Rubin JB, Gibson HD, et al. SDF-1 alpha induces che-motaxis and enhances Sonic hedgehog-induced proliferation ofcerebellar granule cells. Development 2001;128:1971.

30. Peled A, Petit I, Kollet O, et al. Dependence of human stem cellengraftment and repopulation of NOD/SCID mice on CXCR4.Science 1999;283:845.

31. Oh JW, Olman M, Benveniste EN. CXCL12-mediated inductionof plasminogen activator inhibitor-1 expression in humanCXCR4 positive astroglioma cells. Biol Pharm Bull 2009;32:573.

32. Sallusto F, Mackay CR, Lanzavecchia A. The role of chemokinereceptors in primary, effector, and memory immune responses.Annu Rev Immunol 2000;18:593.

33. Ghosh MC, Collins GD, Vandanmagsar B, et al. Activation ofWnt5A signaling is required for CXC chemokine ligand 12-medi-ated T-cell migration. Blood 2009;114:1366.

34. Eash KJ, Means JM, White DW, et al. CXCR4 is a key regulatorof neutrophil release from the bone marrow under basal andstress granulopoiesis conditions. Blood 2009;113:4711.

35. Salogni L, Musso T, Bosisio D, et al. Activin A induces dendriticcell migration through the polarized release of CXC chemokineligands 12 and 14. Blood 2009;113:5848.

36. Seo KS, Park JY, Davis WC, et al. Superantigen-mediated differ-entiation of bovine monocytes into dendritic cells. J Leukoc Biol2009;85:606.

37. Leng Q, Nie Y, Zou Y, et al. Elevated CXCL12 expression in thebone marrow of NOD mice is associated with altered T cell and

stem cell trafficking and diabetes development. BMC Immunol2008;9:51.

38. Xu Q, Yuan X, Xu M, et al. Chemokine CXC receptor 4–mediatedglioma tumor tracking by bone marrow-derived neural progeni-tor/stem cells. Mol Cancer Ther 2009;8:2746.

39. Mirisola V, Zuccarino A, Bachmeier BE, et al. CXCL12/SDF1 ex-pression by breast cancers is an independent prognostic markerof disease-free and overall survival. Eur J Cancer 2009;45:2579.

40. Salmaggi A, Maderna E, Calatozzolo C, et al. CXCL12, CXCR4and CXCR7 expression in brain metastases. Cancer Biol Ther2009;8:1608.

41. Hassan S, Ferrario C, Saragovi U, et al. The influence of tumor-host interactions in the stromal cell-derived factor-1/CXCR4ligand/receptor axis in determining metastatic risk in breastcancer. Am J Pathol 2009;175:66.

42. Welch DR, Goldberg SF. Molecular mechanisms controllinghuman melanoma progression and metastasis. Pathobiology1997;65:311.

43. Sauter ER, Herlyn M. Molecular biology of human melanomadevelopment and progression. Mol Carcinog 1998;23:132.

44. Lu C, Kerbel RS. Interleukin-6 undergoes transition from para-crine growth inhibitor to autocrine stimulator during humanmelanoma progression. J Cell Biol 1993;120:1281.

45. Wu X, Lee VC, Chevalier E, et al. Chemokine receptors astargets for cancer therapy. Curr Pharm Des 2009;15:742.

46. Vandercappellen J, Van Damme J, Struyf S. The role of CXCchemokines and their receptors in cancer. Cancer Lett 2008;267:226.

47. Wendt MK, Johanesen PA, Kang-Decker N, et al. Silencing ofepithelial CXCL12 expression by DNA hypermethylation pro-motes colonic carcinoma metastasis. Oncogene 2006;25:4986.

48. KuciaM, Jankowski K, Reca R, et al. CXCR4-SDF-1 signalling, lo-comotion, chemotaxis and adhesion. J Mol Histol 2004;35:233.