Journal of Surgical Oncology 2008;97:433–438

Expression of CXCL12 and its Receptor CXCR4 Correlates With

Lymph Node Metastasis in Submucosal Esophageal Cancer

KEN SASAKI, MD,* SHOJI NATSUGOE, MD, PhD, SUMIYA ISHIGAMI, MD, PhD, MASATAKAMATSUMOTO, MD, PhD,HIROSHI OKUMURA, MD, PhD, TETSURO SETOYAMA, MD, PhD, YASUTO UCHIKADO, MD, PhD,

YOSHIAKI KITA, MD, KIYOKAZU TAMOTSU, MD, TOSHIHIDE SAKURAI, MD,TETSUHIRO OWAKI, MD, PhD, AND TAKASHI AIKOU, MD, PhD

Department of Surgical Oncology and Digestive Surgery, Field of Oncology, Course of Advanced Therapeutics,Kagoshima University Graduate School of Medical and Dental Sciences, Kagoshima, Japan

Background and Objectives: The chemokine CXCL12 and its receptor CXCR4 are involved in cell migration, proliferation, and angiogenesis,

and promote organ-specific localization of distant metastases in various carcinomas. We examined their expression and microvessel density

(MVD) in submucosal esophageal squamous cell carcinoma (ESCC) and analyzed their connection to clinicopathological findings including

lymph node micrometastasis (LMM).

Methods: Eighty-six patients with submucosal ESCC underwent curative resection from 1985 to 2002. Immunohistochemical staining of

CXCL12, CXCR4, and CD34 was performed with primary tumors, and staining of cytokeratin was performed with dissected lymph nodes. MVD

was calculated from CD34 expression, and LMM detected by cytokeratin staining.

Results: Expression of CXCL12, but not CXCR4, correlated with lymph node metastasis. There was no significant correlation between the

expression of CXCL12 and/or CXCR4 and MVD. LMM was detected in 8 cases and 14 lymph nodes. CXCL12 expression and high MVD were

found in tumors with lymph node metastasis including LMM. Furthermore, in the CXCR4-positive tumors, positive CXCL12 expression was

more significantly correlated with lymph node metastasis and/or LMM than negative CXCL12 expression.

Conclusions: Evaluation of CXCL12 and CXCR4 expression should assist detection of lymph node metastasis including LMM in submucosal

ESCC.

J. Surg. Oncol. 2008;97:433–438. � 2008 Wiley-Liss, Inc.

KEY WORDS: CXCL12; CXCR4; lymph node metastasis; microvessel density; submucosal esophageal squamous cellcarcinoma

INTRODUCTION

Lymph node metastasis is one of the most important prognostic

factors in esophageal squamous cell carcinoma (ESCC). Tumors

sometimes recur even in patients who have undergone complete

resection and have no histological evidence of lymph node metastasis,

which suggests that tumor spread can be more advanced than indicated

by histological diagnosis. The development of sensitive immunohisto-

chemical techniques has allowed detection of lymph node micro-

metastasis (LMM). Cytokeratin (CK), which is a component of some

intermediate filaments, is an epithelial marker and assays of CK

expression have been used to detect individual tumor cells in lymph

nodes of patients with various types of carcinoma [1–4]. We have

investigated the relationship between LMM and clinicopathological

findings in patients with esophageal carcinoma [5–8].

Chemokines are cytokines of 8–10 kDa that are classified into four

groups (CXC, CC, C, and CX3C). Chemokine receptors belong to

the G-protein-coupled receptor superfamily. Chemokines/chemokine

receptors regulate a variety of immune responses to infection, inflam-

mation, and tissue repair. Besides controlling the trafficking of immune

cells, chemokines also regulate the migration of several different cell

types in embryogenesis. Recently, the chemokine CXCL12, also

known as stromal cell-derived factor-1, and its receptor CXCR4 have

been found to have prominent roles in the initiation and progression of

primary and metastatic cancers including breast, ovarian, prostate,

kidney, brain, and lung cancers [9–18].

The aims of this retrospective study were to examine the

relationship between CXCL12 and CXCR4 expression and clinico-

pathological factors, especially lymph node metastasis including

LMM, in surgical specimens of patients with ESCC with submucosal

invasion.

MATERIALS AND METHODS

Patient Characteristics

Eighty-six consecutive patients with ESCC with submucosal

invasion (79 men and 7 women) underwent esophagectomy at

Kagoshima University Hospital from 1985 to 2002. The ages of the

patients ranged from 39 to 81 years (mean 65.8 years). None of the

patients received radiation therapy or chemotherapy before surgery.

Clinicopathological findings were based on the criteria of the TNM

Abbreviations: CK, cytokeratin; ESCC, esophageal squamous-cell carci-noma; LMM, lymph node micrometastasis; MVD, microvessel density;PBS, phosphate-buffered saline.

Grant sponsor: Ministry of Education, Science, Sports and Culture in Japan;Grant number: 17390373.

*Correspondence to: Ken Sasaki, MD, Department of Surgical Oncologyand Digestive Surgery, Field of Oncology, Course of Advanced Therapeu-tics, Kagoshima University Graduate School of Medical and DentalSciences, 8-35-1 Sakuragaoka, Kagoshima 890-8520, Japan. Fax: þ81-99-265-7426. E-mail: k-sasaki@m.kufm.kagoshima-u.ac.jp

Received 10 October 2007; Accepted 5 December 2007

DOI 10.1002/jso.20976

Published online 4 January 2008 in Wiley InterScience(www.interscience.wiley.com).

� 2008 Wiley-Liss, Inc.

classification for esophageal carcinoma of the International

Union Against Cancer [19]. We classified 20 of the ESCCs as

well differentiated, 42 as moderately differentiated, and 24 as poorly

differentiated. Twenty of the tumors were located in the upper third of

the esophagus, 44 in the middle third and 22 in the lower third.

Depth of cancer invasion was histologically classified as follows: sm1,

invasion limited to the upper third of the submucosa (15 patients);

sm2, invasion to between one third and two thirds of the submucosa

(25 patients); and sm3, invasion into the deepest third of the submucosa

(46 patients). Lymph node metastasis was found in 40 of the

86 patients (46.5%). Lymphatic and venous invasion were found

in 55.8% (48/86) and 30.0% (26/86) of the patients, respectively

(Table I).

Immunohistochemical Analysis

Tumor samples were fixed with 10% formaldehyde in phosphate-

buffered saline (PBS), embedded in paraffin, and sectioned into

4-mm-thick slices. They were deparaffinized in xylene and dehydrated

with a series of graded ethanol. For antigen retrieval, sections were

autoclaved in 10 mM citrate buffer solution for 10 min at 1208C for

CXCL12 and CD34 staining or incubated in 0.4% pepsin and 0.2 M

hydrochloric acid for 20 min at 378C for CXCR4 staining. The

endogenous peroxidase activity of specimens was blocked by

immersing the slides in a 3% hydrogen peroxidase-methanol solution

for 30 min at room temperature. After washing three times with PBS

for 5 min each, the sections were treated with 1% bovine serum

albumin for 30 min to block nonspecific reactions at room temperature.

The blocked sections were incubated with primary antibody CXCL12

(R&D Systems, Minneapolis, MN, 1:200), CXCR4 (R&D Systems,

1:200) or CD34 (Transduction Laboratories, Lexington, KY, 1:100)

diluted in PBS at room temperature for 45 min for CXCL12 and

CXCR4 and at 48C for overnight for CD34, followed by staining

with a streptavidin-biotin peroxidase kit (Nichirei, Tokyo, Japan).

The sections were washed three times in PBS for 5 min and the

immune complex was visualized by incubating the sections with

diaminobenzidine tetrahydrochloride. The sections were rinsed briefly

in water, counterstained with hematoxylin, and mounted. Normal

human tonsil and spleen were used as positive controls for CXCL12

and CXCR4, respectively. The primary antibodies were replaced with

PBS for negative controls. Immunohistochemistry was independently

evaluated by two investigators (K.S. and S.I.) who were unaware of the

clinical data or disease outcome. In cases in which the immuno-

histochemical evaluation of the two observers differed, slides were

evaluated by a third observer (S.N.). Positive expression of CXCL12

was defined as detectable immunoreaction in membrane and cyto-

plasmic lesions of >10% of the cancer cells. To evaluate expression of

CXCL12, ten fields (within the tumor and at the invasive front) were

selected and expression in 1,000 tumor cells (100 cells/field) was

evaluated using high-power (200�) microscopy. CXCR4 expression

was categorized into four grades according to homogeneous staining

intensity: absent, weak, moderate, and strong. In the case of hetero-

geneous staining within the sample, more than 50% of the cells were

chosen to determine the staining intensity. We also categorized CXCR4

expression according to intensity: absent is CXCR4 negative; weak to

strong intensity is CXCR4 positive.

For the 46 patients found to be node-negative by conventional

HE staining, one additional section from each lymph node was

immunohistochemically stained using the AE1/AE3 monoclonal anti-

body cocktail (20:1 mixture of AE1 to AE3; Boehringer Mannheim,

Germany), which reacts with a broad spectrum of human CKs [20,21].

All sections were incubated at 608C overnight, deparaffinized in xylene,

and rehydrated with a graded series of ethanol. After heating in citrate

buffer solution (pH 6.0) for 6 min in a pressure cooker, the sections

were incubated with CK monoclonal antibody diluted 1:100. Reactions

for CK were visualized using diaminobenzidine tetrahydrochloride.

The immunohistochemical examination was limited to one slide per

lymph node to expedite routine histological procedures. LMM was

confirmed by clear cytoplasmic immunoreactivity and nuclear shape and

size, compared with the morphology of macrophages, plasma cells,

and reticular cells. In almost all the specimens, detection of a

single metastatic tumor cell was followed by detection of several more

scattered single metastatic tumor cells. In this study, lymph node

micrometastasis was defined as tumor cells with positive cytokeratin

staining, irrespective of the forms of single cell or cluster cells. The

negative controls were sections treated with the same protocol but with

the primary antibody omitted, and sections exposed to normal mouse

IgG instead of the primary antibodies. Normal esophageal epithelium

and the primary tumors of specimens were used as positive controls and

were consistently positive.

Microvessel Quantification

Vessel count was assessed by light microscopy in those areas of the

tumor containing the highest numbers of capillaries and small venules

at the invasive edge. These highly vascular areas were identified by

scanning tumor sections at low power (40� and 100�): six areas were

identified then vessel count was performed in a 200� field (20�objective and 10� ocular), and the average of the six fields was

determined. As described by Weidner et al. [22], identification of a

lumen was not necessary for a structure to be defined as a vessel. For a

high microvessel density (MVD), various cutoffs such as 40, 60, 116,

and 145/mm2 per 200� microscopic field have been used [23]: we

adopted a 35/mm2 cutoff for high MVD based on the median value of

the whole group.

Statistical Analysis

The Pearson chi-square test was used to compare qualitative variables

and the quantitative variables were analyzed by non-parametric testing

Journal of Surgical Oncology

TABLE I. Clinicopathological Characteristics of 86 Patients With

Submucosal Esophageal Squamous Cell Carcinoma

Features n (%)

Mean age� SD, years 65.8� 8.2

Gender

Male 79 (91.9)

Female 7 (8.1)

Location of tumor

Upper 20 (23.2)

Middle 44 (51.2)

Lower 22 (25.6)

Histopathologic grading

Well 20 (23.3)

Moderate 42 (48.9)

Poor 24 (27.9)

Depth of tumor invasion

sm1 15 (17.4)

sm2 25 (29.1)

sm3 46 (53.5)

Lymphatic invasion

Negative 38 (44.2)

Positive 48 (55.8)

Venous invasion

Negative 60 (70.0)

Positive 26 (30.0)

Lymph node metastasis

Negative 46 (53.5)

Positive 40 (46.5)

SD, standard deviation.

434 Sasaki et al.

(Wilcoxon test). P< 0.05 was considered to be statistically significant.

All statistical analyses used the software package JMP 5 for Windows

software (SAS Institute, Inc., Cary, NC).

RESULTS

CXCL12 and CXCR4 Expression and MVD in

Submucosal Esophageal Squamous Cell Carcinoma

Expression of CXCL12 and CXCR4 was detected in both

the membrane and the cytoplasm of ESCC (Fig. 1A,B). Expression

of CXCL12 and CXCR4 was observed in 32.6% (28/86) and 87.2%

(75/86) of patients, respectively. CD34 expression was detected in both

blood and lymphatic endothelial cells (Fig. 1C). The mean MVD was

39.1� 22.7 (range 8–105/field).

Relationship Between CXCL12 and CXCR4 Expression

and Clinicopathological Findings Including MVD

Expression of CXCL12 was associated with lymph node metastasis

(P¼ 0.0212). Expression of CXCR4 had no correlation with clinico-

pathological variables or the expression of CXCL12. We wondered

whether CXCL12 and CXCR4 expression was more frequent in patients

with a higher MVD: no significant relationship emerged (Table II).

LMM and Correlation With CXCL12 and

CXCR4 Expression and MVD

Fourteen LMM in 8 of the 46 patients shown to be node-negative by

conventional HE staining were detected by CK immunostaining

(Fig. 2A,B). CXCL12 expression and high MVD correlated signi-

ficantly with lymph node metastasis including LMM (P¼ 0.0112 and

P¼ 0.0024, respectively; Table III). Interestingly, although CXCR4

expression was not related to nodal metastasis including LMM, all

patients with LMM had positive CXCR4 expression. Furthermore,

expression of both CXCL12 and CXCR4 correlated more significantly

with lymph node metastasis including LMM than CXCL12 expression

only (P¼ 0.0074; Table III). When dividing the patients without nodal

metastasis into two groups by histological examination, the presence or

absence of LMM and positive rates of CXCL12 and MVD were higher

for patients with LMM than for those without LMM.

DISCUSSION

Many studies of the CXCR4/CXCL12 pathway have been reported,

mainly in the field of immunology and infection, in areas such as

hematopoiesis, lymphocyte homing, and HIV infection. In 2001, Muller

et al. showed that CXCR4 was expressed more highly in breast cancer

tissue than in normal breast tissue and that CXCL12 was expressed in

many organs, such as lymph nodes, bone marrow, and lungs, in which

breast cancer metastasis is often found, but not in kidneys, where

metastasis hardly occurs. In addition, in vivo studies showed that

metastasis to bone marrow and lungs was inhibited by injecting antibody

that neutralizes CXCR4 activity [10]. Subsequently, interest grew in the

role of the CXCR4/CXCL12 pathway in tumors.

In the present study, we immunohistochemically examined the

expression of CXCL12 and CXCR4 in the primary tumor of the

esophagus. All patients with LMM had positive CXCR4 expression in

the primary tumor, although the incidence of negative expression of

CXCR4 was low. Kaifi et al. [24] reported that CXCR4 might have a

role in early metastatic spread because its expression is associated

with micrometastasis to both lymph nodes and bone marrow. We

have previously found that patients with LMM with ESCC or gastric

adenocarcinoma had significantly poorer prognoses and greater in-

cidence of postoperative recurrence than those without LMM [6–8].

Journal of Surgical Oncology

Fig. 1. Immunohistochemical staining of CXCL12, CXCR4, and CD34 in esophageal squamous cell carcinoma with submucosal invasion.A: Positive expression of CXCL12. B: Positive expression of CXCR4. C: Intratumoral blood and lymphatic microvessels were detected as singleor clustered endothelial cells with or without lumen (counterstain: Mayer’s hematoxylin; original magnification, 200�).

CXCL12 and CXCR4 in Esophageal Cancer 435

Thus, detection of LMM seems to be important for selection of

treatment and prediction of clinical outcome.

However, few studies have focused on protein expression of

CXCL12 and CXCR4 in cancer cells and a relationship between LMM

and MVD. There have been no reports disclosing the clinical

implications of chemokine positivity in tumor cells, except for

some reports on CXCL12 expression, which significantly correlated

with prognosis in glioma cells [25] and gastric adenocarcinoma cells

[26], but not in epithelial ovarian cancer cells [27] or oral squamous

cell carcinoma [28]. Although the relationship between CXCR4

expression and malignant potential has been investigated in

various types of cancer, its clinical significance is still controversial

[24,28–30].

Here, for the first time to our knowledge, we examined patients

with ESCC with submucosal invasion, excluding advanced cases,

to investigate a correlation between clinicopathological factors,

especially lymph node metastasis, and expression of CXCL12 and

CXCR4. Our results indicate a close correlation between CXCL12

expression and lymph node metastasis including LMM, which occurs

at an early stage. Furthermore, CXCL12 expression in CXCR4-

positive tumors was more strongly related to lymph node metastasis

including LMM. Therefore, expression of either chemokine or its

receptor might not be important for the regulation of malignant

properties, but the expression of both chemokine and receptor together

may be important for lymph node metastasis including LMM.

During normal development, the chemokine CXCR4 is highly

expressed in the endothelium of developing blood vessels of the

gastrointestinal tract [31]. Furthermore, mice lacking either the

chemokine CXCL12 or its receptor CXCR4 have defective vascular

development, particularly in the gastrointestinal tract, suggesting that

CXCL12 and CXCR4 represent an important signaling system for

organ vascularization [31]. The CXCL12/CXCR4 pathway regulates

the expression and secretion of the glycolytic enzyme phosphoglyce-

rate kinase 1, an important regulator of angiogenesis in prostate cancer

[32]. However, we found no correlation between the expression of

CXCL12 and CXCR4 and MVD. Further studies are needed to clarify

the correlation between angiogenesis and the CXCL12/CXCR4

pathway in cancer cells.

Journal of Surgical Oncology

TABLE II. CXCL12 and CXCR4 Expression in Relation to Clinicopathological Findings in Submucosal Esophageal Squamous Cell Carcinomas

Total (n¼ 86)

CXCL12 expression

P-value

CXCR4 expression

P-value

Positive Negative Positive Negative

n¼ 28 (32.6%) n¼ 58 (67.4%) n¼ 75 (87.2%) n¼ 11 (12.8%)

Histopathologic grading

Well 20 6 (30.0) 14 (70.0) NSa 18 (90.0) 2 (10.0) NSa

Moderate 42 13 (31.0) 29 (69.0) 37 (88.1) 5 (11.9)

Poor 24 9 (37.5) 15 (62.5) 20 (83.3) 4 (16.7)

Depth of tumor invasion

sm1 15 3 (20.0) 12 (80.0) NSa 12 (80.0) 3 (20.0) NSa

sm2 25 6 (24.0) 19 (76.0) 22 (88.0) 3 (12.0)

sm3 46 19 (41.3) 27 (58.7) 41 (89.1) 5 (10.9)

pN

pN0 46 10 (21.7) 36 (78.3) 0.021a 42 (91.3) 4 (8.7) NSa

pN1 40 18 (45.0) 22 (55.0) 33 (82.5) 7 (17.5)

Lymphatic invasion

Negative 38 9 (23.7) 29 (76.3) NSa 33 (86.8) 5 (13.2) NSa

Positive 48 19 (39.6) 29 (60.4) 42 (87.5) 6 (12.5)

Venous invasion

Negative 60 19 (31.7) 41 (68.3) NSa 51 (85.0) 9 (15.0) NSa

Positive 26 9 (34.6) 17 (65.4) 24 (92.3) 2 (7.7)

Mean MVD� SD 40.1� 22.0 38.2� 23.1 NSb 38.7� 22.0 42.1� 28.3 NSb

MVD, microvessel density; SD, standard deviation; NS, not significant.aStatistical analyses by chi-square test.bStatistical analysis by Wilcoxon test.

Fig. 2. Immunohistochemical staining of AE1/AE3 in lymph node found to be negative for metastasis by conventional hematoxylin-eosinstaining. A: Single-cell micrometastasis. B: Cluster-type micrometastasis. (counterstain: Mayer’s hematoxylin; original magnification, 400�).

436 Sasaki et al.

Our results are consistent with the hypothesis that activation of the

CXCL12/CXCR4 pathway could promote lymph node metastasis in

submucosal ESCC. This pathway could be targeted by blocking the

expression of CXCR4 in tumor cells or the secretion of CXCL12 by

tumor cells and stromal cells in the tumor microenvironment. Various

blocking agents against CXCR4 have been developed [33,34]; they

should be used to validate this chemokine receptor as a molecular

target for the treatment of ESCC patients who have not received toxic

systemic treatment.

In conclusion, we found that CXCL12 expression, especially in

expression of both CXCL12 and CXCR4, correlated with lymph

node metastasis including LMM in submucosal ESCC. The relation-

ship between CXCL12 and CXCR4 expression and MVD was not

significant suggesting that CXCL12/CXCR4 was much less likely

to associate with angiogenesis in submucosal ESCC. While such

chemokines might have a role as markers of nodal metastasis or even as

mediators for tumor spread, our data suggest they exert their effects by

mechanisms other than that of mediating lymphatic invasion. Further

study should be necessary in the mechanism between CXCL12 and

CXCR4 expression and nodal metastasis.

ACKNOWLEDGMENTS

This study was supported by Grant-in-Aid 17390373 for Scientific

Research from the Ministry of Education, Science, Sports and Culture

in Japan.

REFERENCES

1. Maehara Y, Oshiro T, Endo K, et al.: Clinical significance ofoccult micrometastasis lymph nodes from patients with earlygastric cancer who died of recurrence. Surgery 1996;119:397–402.

2. Greenson JK, Isenhart CE, Rice R, et al.: Identification of occultmicrometastases in pericolic lymph nodes of Duke’s B colorectalcancer patients using monoclonal antibodies against cytokeratinand CC49. Correlation with long-term survival. Cancer 1994;73:563–569.

3. Cutait R, Alves VA, Lopes LC, et al.: Restaging of colorectalcancer based on the identification of lymph node micrometastasesthrough immunoperoxidase staining of CEA and cytokeratins. DisColon Rectum 1991;34:917–920.

4. Clare SE, Sener SF, Wilkens W, et al.: Prognostic significance ofoccult lymph node metastases in node-negative breast cancer. AnnSurg Oncol 1997;4:447–451.

5. Matsumoto M, Natsugoe S, Okumura H, et al.: Overexpression ofvascular endothelial growth factor-C correlates with lymph nodemicrometastasis in submucosal esophageal cancer. J GastrointestSurg 2006;10:1016–1022.

6. Matsumoto M, Natsugoe S, Ishigami S, et al.: Rapid immuno-histochemical detection of lymph node micrometastasis duringoperation for upper gastrointestinal carcinoma. Br J Surg 2003;90:563–566.

7. Matsumoto M, Natsugoe S, Nakashima S, et al.: Clinicalsignificance of lymph node micrometastasis of pN0 esophagealsquamous cell carcinoma. Cancer Lett 2000;153:189–197.

8. Natsugoe S, Mueller J, Stein HJ, et al.: Micrometastasis and tumorcell microinvolvement of lymph nodes from esophageal squ-amous cell carcinoma: Frequency, associated tumor character-istics, and impact on prognosis. Cancer 1998;83:858–866.

9. Scotton CJ, Wilson JL, Milliken D, et al.: Epithelial cancer cellmigration: A role for chemokine receptors? Cancer Res 2001;61:4961–4965.

10. Muller A, Homey B, Soto H, et al.: Involvement of chemokinereceptors in breast cancer metastasis. Nature 2001;410:50–56.

11. Kijima T, Maulik G, Ma PC, et al.: Regulation of cellularproliferation, cytoskeletal function, and signal transductionthrough CXCR4 and c-Kit in small cell lung cancer cells. CancerRes 2002;62:6304–6311.

12. Zhou Y, Larsen PH, Hao C, et al.: CXCR4 is a major chemokinereceptor on glioma cells and mediates their survival. J Biol Chem2002;277:49481–49487.

13. Schrader AJ, Lechner O, Templin M, et al.: CXCR4/CXCL12expression and signalling in kidney cancer. Br J Cancer 2002;86:1250–1256.

14. Taichman RS, Cooper C, Keller ET, et al.: Use of the stromal cell-derived factor-1/CXCR4 pathway in prostate cancer metastasis tobone. Cancer Res 2002;62:1832–1837.

15. Uchida D, Begum NM, Almofti A, et al.: Possible role of stromal-cell-derived factor-1/CXCR4 signaling on lymph node metastasisof oral squamous cell carcinoma. Exp Cell Res 2003;290:289–302.

16. Hwang JH, Hwang JH, Chung HK, et al.: CXC chemokinereceptor 4 expression and function in human anaplasticthyroid cancer cells. J Clin Endocrinol Metab 2003;88:408–416.

17. Phillips RJ, Burdick MD, Lutz M, et al.: The stromal derivedfactor-1/CXCL12-CXC chemokine receptor 4 biological axis innon-small cell lung cancer metastases. Am J Respir Crit Care Med2003;167:1676–1686.

18. Mori T, Doi R, Koizumi M, et al.: CXCR4 antagonist inhibitsstromal cell-derived factor 1-induced migration and invasionof human pancreatic cancer. Mol Cancer Ther 2004;3:29–37.

19. Sobin LH, Fleming ID: TNM Classification of Malignant Tumors,fifth edition (1997) Union Internationale Contre le Cancer andthe American Joint Committee on Cancer. Cancer 1997;80:1803–1804.

20. Listrom MB, Dalton LW: Comparison of keratin monoclonalantibodies MAK-6, A E1:AE3, and CAM-5.2. Am J Clin Pathol1987;88:297–301.

21. Woodcock-Mitchell J, Eichner R, Nelson WG, et al.: Immuno-localization of keratin polypeptides in human epidermis usingmonoclonal antibodies. J Cell Biol 1982;95:580–588.

22. Weidner N, Semple JP, Welch WR, et al.: Tumor angiogenesisand metastasis—Correlation in invasive breast carcinoma. N EnglJ Med 1991;324:1–8.

23. Choi JY, Jang KT, Shim YM, et al.: Prognostic significance ofvascular endothelial growth factor expression and microvesseldensity in esophageal squamous cell carcinoma: Comparison withpositron emission tomography. Ann Surg Oncol 2006;13:1054–1062.

24. Kaifi JT, Yekebas EF, Schurr P, et al.: Tumor-cell homing tolymph nodes and bone marrow and CXCR4 expression inesophageal cancer. J Natl Cancer Inst 2005;97:1840–1847.

25. Salmaggi A, Gelati M, Pollo B, et al.: CXCL12 expression ispredictive of a shorter time to tumor progression in low-grade

Journal of Surgical Oncology

TABLE III. Correlation Between Expression of CXCL12, CXCR4, and

Microvessel Density According to Lymph Node Metastasis Including

Micrometastasis

Factor

pN(�) and

LMM(�)

pN(þ) and/or

LMM(þ) P-value

CXCL12

Negative 31 27 0.0112a

Positive 7 21

CXCR4

Negative 4 7 NSa

Positive 34 41

Mean MVD� SD 31.5� 20.8 45.2� 22.5 0.0024b

CXCR4 positive group

CXCL12

Negative 28 22 0.0074a

Positive 6 19

CXCR4 negative group

CXCL12

Negative 3 5 NSa

Positive 1 2

LMM, lymph node micrometastasis; MVD, microvessel density; SD, standard

deviation; NS, not significant.aStatistical analyses by chi-square test.bStatistical analysis by Wilcoxon test.

CXCL12 and CXCR4 in Esophageal Cancer 437

glioma: A single-institution study in 50 patients. J Neurooncol2005;74:287–293.

26. Ishigami S, Natsugoe S, Okumura H, et al.: Clinical implicationof CXCL12 expression in gastric cancer. Ann Surg Oncol 2007;14:3154–3158.

27. Jiang YP, Wu XH, Shi B, et al.: Expression of chemokineCXCL12 and its receptor CXCR4 in human epithelial ovariancancer: An independent prognostic factor for tumor progression.Gynecol Oncol 2006;103:226–233.

28. Almofti A, Uchida D, Begum NM, et al.: The clinicopathologicalsignificance of the expression of CXCR4 protein in oral squamouscell carcinoma. Int J Oncol 2004;25:65–71.

29. Laverdiere C, Hoang BH, Yang R, et al.: Messenger RNAexpression levels of CXCR4 correlate with metastatic behaviorand outcome in patients with osteosarcoma. Clin Cancer Res2005;11:2561–2567.

30. Schimanski CC, Schwald S, Simiantonaki N, et al.: Effect ofchemokine receptors CXCR4 and CCR7 on the metastaticbehavior of human colorectal cancer. Clin Cancer Res 2005;11:1743–1750.

31. Tachibana K, Hirota S, Iizasa H, et al.: The chemokine receptorCXCR4 is essential for vascularization of the gastrointestinaltract. Nature 1998;393:591–594.

32. Wang J, Wang J, Dai J, et al.: A glycolytic mechanism regulatingan angiogenic switch in prostate cancer. Cancer Res 2007;67:149–159.

33. Liang Z, Wu T, Lou H, et al.: Inhibition of breast cancermetastasis by selective synthetic polypeptide against CX CR4.Cancer Res 2004;64:4302–4308.

34. Tamamura H, Hori A, Kanzaki N, et al.: T140 analogs as CXCR4antagonists identified as anti-metastatic agents in the treatment ofbreast cancer. FEBS Lett 2003;550:79–83.

Journal of Surgical Oncology

438 Sasaki et al.