Tumor lymphangiogenesis correlates with lymph node metastasis and clinicopathologic parameters in oral squamous cell carcinoma


  • Mayumi Miyahara DDS,

    1. Department of Maxillofacial Diagnostic and Surgical Science, Field of Oral and Maxillofacial Rehabilitation, Kagoshima University Graduate School of Medical and Dental Sciences, Kagoshima, Japan
    Search for more papers by this author
  • Jun-ichi Tanuma DDS, PhD,

    Corresponding author
    1. Department of Oral Pathology, Field of Oncology, Kagoshima University Graduate School of Medical and Dental Sciences, Kagoshima, Japan
    • Department of Oral Pathology, Field of Oncology, Kagoshima University Graduate School of Medical and Dental Sciences, Kagoshima 890-8544, Japan
    Search for more papers by this author
    • Fax: (011) 81-99-275-6148

  • Kazumasa Sugihara DDS, PhD,

    1. Department of Maxillofacial Diagnostic and Surgical Science, Field of Oral and Maxillofacial Rehabilitation, Kagoshima University Graduate School of Medical and Dental Sciences, Kagoshima, Japan
    Search for more papers by this author
  • Ichiro Semba DDS, PhD

    1. Department of Oral Pathology, Field of Oncology, Kagoshima University Graduate School of Medical and Dental Sciences, Kagoshima, Japan
    Search for more papers by this author



Lymphatic vessel density (LVD) and microvessel density (MVD) are important parameters for assessing the malignant potential of tumors and patient survival. In this report, the authors defined LVD as the density of D2-40-positive lymphatic vessels and MVD as the density of CD105-positive microvessels per unit area of tissue. It was reported previously that vascular endothelial growth factor C (VEGF-C) is a major modulator of LVD and MVD. The objectives of this study were to clarify the clinical and prognostic significance of both LVD and MVD in oral squamous cell carcinoma (OSCC) and to elucidate the lymphangiogenic and angiogenic activities of VEGF-C in cancer tissues.


In total, 110 OSCC tissue samples were evaluated for LVD, MVD, and expression of VEGF-C using immunohistochemistry. Correlations among these parameters and clinicopathologic factors were examined.


LVD was significantly higher in tumors that had very high expression of VEGF-C compared with tumors that had no/weak expression of VEGF-C. LVD correlated well with lymph node metastasis (P < .001). MVD was correlated significantly with positive lymph node metastasis (P < .001) but not with VEGF-C expression. In contrast, high expression of VEGF-C was correlated significantly with advanced tumor status (P = .041). Survival rates were lower in patients who had higher LVD (P < .001), higher MVD (P = .0028), and strong VEGF-C expression (P = .048).


Lymphangiogenesis predominantly influenced metastasis-free survival. The current results suggested that LVD is a more useful tool than MVD and VEGF-C for deciding on therapeutic strategies in patients with OSCC. Cancer 2007. © 2007 American Cancer Society.

Metastatic dissemination of a primary tumor is an important factor that has a negative effects on the prognosis for patients with malignancies. Tumor angiogenesis plays a critical role in the growth and systemic dissemination of malignancies.1 In addition to the route through blood vessels, the lymphatic system also contributes to tumor cell dissemination. Indeed, the spread to regional lymph nodes is an early event in systemic dissemination. However, the clinical significance of the de novo formation of lymphatic capillaries (lymphangiogenesis) and its regulation in various cancers remain unclear, largely because specific markers for lymphatic vessel endothelium remain unknown, making it difficult to discriminate between lymphatic vessels and blood capillaries in cancer tissues. Recently, several new, specific antibodies for lymphatic endothelial cells have been developed and applied to investigate lymphangiogenesis in human cancers.2

The D2-40 antibody detects a fixation-resistant epitope on podoplanin, which is a selective marker for lymphatic endothelium, allowing the identification of lymphatic vessels in formalin-fixed, paraffin-embedded tissues.3 The Flt-4 antibody recognizes the vascular endothelial growth factor (VEGF) receptor 3 (VEGFR-3), a tyrosine kinase receptor that is expressed on lymphatic endothelium. VEGF-C-activated VEGFR-3 induces the growth of both lymphatic vessels and blood vessels and promotes the proliferation, migration, and survival of cultured human adult lymphatic endothelium.4 Overexpression of VEGF-C increases Flt-4-positive vessel density (FVD) and accelerates lymphatic invasion or lymph node metastasis in experimental tumors5 and in a variety of human cancers.6

Oral squamous cell carcinoma (OSCC) is the most common malignant tumor of the head and neck region. Head and neck squamous cell carcinomas (SCCs) preferentially spread through lymphatic vessels, and the presence of lymph node metastasis at the time of diagnosis is an indication of poor prognosis.7, 8 Among various determinants of tumor prognosis, tumor microvessel density (MVD) is one of the most useful factors for predicting tumor growth, grade of tumor differentiation, recurrence-free survival, and overall survival. However, to our knowledge, the clinical and pathologic significance of lymphangiogenesis revealed by lymphatic vessel density (LVD) has not been investigated in human OSCC. For this study, we evaluated LVD, MVD, and expression levels of VEGF-C, Flt-4, and Ki-67 in OSCC using immunohistochemistry to clarify their significance in the proliferation of lymphatic vessels and to correlate them with the presence of lymph node metastasis and clinical outcomes in patients with OSCC.



OSCC biopsy specimens from 110 patients were obtained from the Department of Oral and Maxillofacial Surgery, Kagoshima University Hospital. The patients who were involved underwent surgery at our institution from 1992 to 2001 but received no other prior treatment. Informed consent was obtained from all patients prior to surgery. Patients' records were reviewed and are summarized in Table 1. Tumor (T) classification of OSCC was established using the International Union Against Cancer classification system,9 and histologic grade was allocated according to World Health Organization protocols.10

Table 1. Correlation Between Density of Lymphatics and Blood Vessels and Clinicopathologic Features of 110 Patients With Oral Squamous Cell Carcinoma
CharacteristicNo. of patients (%)Mean ± SDVEGF-C expression
  1. SD indicates standard deviation; VEGF-C, vascular endothelial growth factor C; LVD, lymphatic vessel density; MVD. microvessel density; FVD, Flt-4-positive vessel density; mitotic index, expression of Ki-67.

Age, y
 <6553 (48.2)20.8 ± 10.621.2 ± 12.83.23 ± 4.1718 35
 >6557 (51.8)23.2 ± 16.222.2 ± 20.13.98 ± 3.0612 45
 P .265.542.129 .128 
 Men75 (68.2)29.6 ± 16.524.3 ± 8.013.93 ± 4.6120 55
 Women35 (31.8)27.1 ± 16.621.6 ± 10.93.43 ± 2.3710 25
 P .312.256.211 .828 
Tumor site
 Tongue46 (41.8)28.3 ± 12.523.3 ± 10.24.75 ± 4.4318 28
 Other region64 (58.2)16.5 ± 5.6615.4 ± 13.13.19 ± 2.7812 52
 P <.001<.001.011 .029 
Tumor classification
 T1/T283 (75.5)18.7 ± 3.6122.2 ± 2.793.73 ± 2.3228 55
 T3/T427 (24.5)30.9 ± 3.7025.9 ± 2.413.85 ± 2.722 25
 P <.001.159.895 .006 
Histologic grade
 Well differentiated71 (64.5)23.3 ± 4.9820.8 ± 3.943.67 ± 3.2424 47
 Moderately/poorly differentiated39 (35.5)22.5 ± 5.0520.5 ± 4.663.54 ± 2.486 33
 P .323.699.845 .045 
Lymph node involvement
 Positive31 (28.2)34.8 ± 2.5133.1 ± 4.344.43 ± 2.5310 21
 Negative79 (71.8)14.9 ± 2.8618.2 ± 3.393.43 ± 1.9820 59
 P <.001<.001.012 .482 
Mode of invasion
 Grade 1/217 (15.4)11.7 ± 1.8615.2 ± 2.482.85 ± 1.166 11
 Grade 3/493 (84.6)30.6 ± 4.8326.4 ± 4.564.11 ± 2.8922 71
 P <.001<.001<.001 .365 

All specimens were fixed routinely in 10% formalin and embedded in paraffin. Serial 3- or 4-μm sections were cut from paraffin blocks. One section from each tumor was stained with hematoxylin and eosin to reevaluate the histologic diagnosis, and the other sections were used for immunohistochemistry.


Using the antibodies listed in Table 2, immunohistochemical staining was performed with the Envision+ Dual Link/horseradish peroxidase system (Dako, Glostrup, Denmark). Dewaxed sections were pretreated as described in Table 2 for epitope retrieval. Endogenous peroxidase activity was blocked with 3% hydrogen peroxide for 10 minutes at room temperature. Slides were then washed with Tris-buffered saline (TBS) and were incubated with either anti-Ki-67, anti-CD105, anti-D2-40, anti-VEGF-C, or anti-Flt-4 antibody overnight at 4°C. Commercial sources and dilutions of each antibody are listed in Table 2. After incubation with the primary antibodies, the sections were rinsed in TBS and were incubated with the appropriate secondary antibodies (antirabbit or antimouse immunoglobulins) conjugated with peroxidase-labeled dextran polymers for 30 minutes at room temperature. After rinsing in TBS, sections were treated with a 0.5 mg/mL 3,3′-diaminobenzidine solution, which contained 0.001% hydrogen peroxide to visualize reaction products. After brief counterstaining with Mayer hematoxylin, sections were dehydrated and mounted.

Table 2. Antibodies Used and Dilutions
  1. VEGF-C indicates vascular endothelial growth factor C.

D2-40D2-40PredilutedNoneSignet, Dedham, Mass
CD105SN6h×5Proteinase KDako, Glostrup, Denmark
VEGF-CRabbit, polyclonal×100Microwave, proteinase KZymed, San Francisco, Calif
Flt-4Rabbit, polyclonalPredilutedTrypsinSpring Bioscience, Fremont, Calif

Quantification of MVD, LVD, and FVD

The densities of the 3 types of vessels (ie, MVD, LVD, and FVD) were evaluated according to the methods described by Weidner et al.11 First, all slides were screened using a low-magnification objective lens to identify the areas that contained the highest number of positively stained vessels (hot spots). The number of vessels was counted in 3 fields in the hot spots using a ×200 magnification lens. Any single, brown-stained cell or cluster of endothelial cells that clearly was separated from the adjacent microvessels, tumor cells, and other connective tissue elements was considered positive for the antibodies that were used. The mean number of vessels from the 3 fields was converted to the number of vessels per mm2 area and was defined as LVD, MVD, or FVD. To determine the relation between these variables and clinical outcome, tissue samples were divided into 2 groups according to density (ie, those with a value higher than the median value for the entire group and those with a value lower than the median).7 Median values were compared using the Kruskal-Wallis 1-way analysis of variance by ranks followed by the Dunn multiple comparison test, when appropriate. High versus weak immunostaining in carcinomas was compared using the chi-square test.

Ki-67 and VEGF-C Expression

For each section that was stained with anti-Ki-67 and anti-VEGF-C antibodies, ≥500 tumor cells were counted using a ×200 magnification objective lens; and the percentage of positively stained cells was determined. A percentage of positively stained tumor cells <30% was scored as low, a percentage of from 30% to 60% was considered moderate, and a percentage >60% was referred to as high.12

Statistical Analysis

Differences in continuous variables for patients' clinical and pathologic characteristics were assessed using a Wilcoxon rank-sum test. A chi-square test or a Fisher exact test was used to assess the associations among categorical data. Clinical and pathologic characteristics were analyzed with survival using Cox proportional-hazards models. Estimates of survival curves were calculated according to the Kaplan-Meier product-limit method. The length of survival among patients with different characteristics was compared using the log-rank test. All computations were performed with SYSTAT 11 (SYSTAT Software, Inc., Calif) and the SPSS software package (version 10.1; SPSS Inc., Chicago, Ill) on a Dell computer with the Windows XP operating system. All P values <.05 were considered statistically significant.



Consistent with a previous report,3 D2-40 (Fig. 1A) was restricted to the endothelium not only at the peritumoral lesion but also at the intratumoral lesion, indicating that D2-40-positive vessels did not appear to form any specific distribution pattern. Table 1 shows that higher LVD values were correlated significantly with higher scores of T classification (P < .001), lymph node involvement (P < .001), and mode of invasion (P < .001). Cancers of the tongue had significantly higher LVD than tumors at other oral sites (P < .001). However, no correlation was observed between LVD and age, sex, or histologic grade.

Figure 1.

(A) Lymphatic endothelial cells specifically were stained intensely with the D2-40 antibody, whereas blood vessel endothelial cells were not stained. (B) The endothelial cells from microvessels specifically were stained intensely with the CD105 antibody. (C) Lymphatic endothelial cells and the peritumoral lesion (cytoplasm) from oral squamous cell carcinoma (OSCC) specifically were stained intensely with the Flt-4 antibody. (D) Staining of lymphatic vessels in OSCC using the D2-40 antibody. (E) Staining of lymphatic vessels in OSCC with the Flt-4 antibody.


Table 1 summarizes the correlations between MVD (Fig. 1B) and clinicopathologic variables. A higher MVD value was correlated positively with higher scores of lymph node involvement (P < .001), and mode of invasion (P < .001) but not with age, sex, T classification, or histologic grade. Cancers of the tongue also had a significantly higher MVD than tumors at other oral sites (P < .001).


The Flt-4-positive vessels (Fig. 1C) also were more or less positive for D2-40 (Fig. 1D). However, strongly positive vessels differed somewhat between these 2 types of vessels. Flt-4 was present in the lymphatic endothelium and in the vascular endothelium. D2-40 was present only in the lymphatic endothelium.

Table 1 shows that a higher FVD value was correlated with the mode of invasion (P < .001) and somewhat with lymph node involvement (P = .012). There was no significant association of FVD with age, sex, T classification, or histologic grade. Among the current group of patients with OSCC, tongue cancers tended to have a higher FVD values than tumors of other oral sites (P = .011).

Expression of VEGF-C

The expression of VEGF-C varied among the OSCC samples; however, the majority of tumors showed at least focal VEGF-C staining, except for 1 tumor without any signal. Table 1 shows that VEGF-C expression by tumor cells was not correlated with patient age, sex, tumor site, histologic grade, mode of invasion, or lymph node involvement. The only exception was greater VEGF-C expression in tumors with higher T classification (P = .006). Table 3 shows that there was no correlation between VEGF-C expression and LVD, MVD, FVD, or the mitotic index of cancer cells, as determined according to the percentages of Ki-67-positive cells.

Table 3. Correlation Between the Expression of Vascular Endothelial Growth Factor C and D2-40, CD105, Flt-4, and the Ki-67 Index in Oral Squamous Cell Carcinoma
VEGF-C expressionMean ± SD
LVD: D2-40MVD: CD105FVD: Flt-4Mitotic index, %
  1. SD indicates standard deviation; VEGF-C, vascular endothelial growth factor C; LVD, lymphatic vessel density; MVD, microvessel density; FVD, Flt-4-positive vessel density; mitotic index, expression of Ki-67.

Low/moderate18.18 ± 6.0120.12 ± 7.144.05 ± 1.724.05 ± 1.72
High23.36 ± 8.2122.02 ± 6.014.28 ± 1.914.05 ± 1.72

Survival Analysis

In a univariate analysis, high LVD (P < .001) (Fig. 2A), high MVD (P = .0028) (Fig. 2B), and higher VEGF-C expression (P = .048; data not shown) were associated with poor survival. FVD did not influence survival (P = .069; data not shown).

Figure 2.

(A) Survival rates among patients with oral squamous cell carcinoma (OSCC) who had high and low lymphatic vessel density (LVD) values (Kaplan-Meier plot; P < .001). (B) Survival rates among patients with OSCC who had high and low microvessel density (MVD) values (Kaplan-Meier plot; P = .0028).


Malignant tumors invade the lymphatics through either intratumoral lymphatic vessels, preexisting lymphatics located at the tumor periphery, or newly generated lymphatics induced by the tumor cells themselves.13 Their relative importance remains unclear, and they may vary according to the types of cancer. Recent increasing evidence favors the possibility that tumor cells actively contribute to lymphatic dissemination by inducing lymphangiogenesis.3, 14

The difficulty in discriminating lymphatics from blood capillaries in formalin-fixed, paraffin-embedded tissues has been overcome by the advent of the new monoclonal antibody D2-40, which recognizes an onocofetal membrane antigen designated M2A: M2A is a heavily O-glycosylated sialoprotein that was found first in testicular and extratesticular germ cell tumors.15, 16 The D2-40 antibody specifically and intensely labels lymphatic endothelial cells, as reported previously, highlighting the presence of lymphatic invasion in tumors.17, 18 In contrast, the blood vessel endothelium is negative for D2-40. Therefore, D2-40 is a new, selective marker for lymphatic endothelium that is applicable for the detection of lymphatic invasion of various malignant neoplasms, including OSCC.17, 19

Increased MVD values reportedly are associated with metastasis, advanced tumor stage, and a poor prognosis in many kinds of malignancies.20–22 However, in head and neck tumors, its clinical significance is controversial.23 Discrepancies in the effects of MVD on clinical variables may reflect differences in treatment protocols and or in the endothelial markers used for immunohistochemical staining. Among commonly used antibodies to stain vascular endothelium, such as anti-von Willebrand Factor VIII, CD 31, and CD 34, the antiendoglins (CD105) are tools that are used to assess tumor vasculature, mainly because CD105 expression is associated with endothelial cell proliferation.7, 24 Although it has been demonstrated that CD105 expression is relevant to tumor metastasis in various malignancies, only a few studies have investigated CD105 expression in cancers of the oral region.23, 25 Our current data showed that MVD was correlated positively with lymph node involvement, mode of invasion, and tumor site. These findings imply that high MVD values also predict more aggressive tumor behavior in OSCC.

Flt-4 is a receptor for lymphangiogenic VEGF-C and VEGF-D and is activated specifically by VEGF-C and VEGF-D. Flt-4 is expressed almost exclusively in the lymphatic endothelium and plays an important role in the development and maintenance of lymphatic vessels26 as a major regulator of lymphagiogenesis in normal tissues. In different types of malignant tumors and granulation tissues, Flt-4 also is expressed on blood vessels.27 Flt-4 expression may provide a new microvascular progression marker, and its presence in tumor blood and lymphatic vessels suggests that it may be a mediator of neovascularization induced by lymphangiogenic factors. A strong, positive correlation was reported between the expression levels of VEGF-C and Flt-4 in breast cancer.6 However, in the current study, FVD was not associated with any clinicopathologic parameters or with VEGF-C expression in OSCC.

Recent studies have revealed the clinicopathologic significance of VEGF-C in various malignancies.12, 28, 29 Positive correlations between lymph node metastasis or lymphatic vessel invasion and VEGF-C expression have been reported in patients with a variety of carcinomas.28, 30, 31 In gastric and cervical carcinomas, VEGF-C has been the only prognostic predictor.28 In contrast to other reports,32 no correlation between VEGF-C expression and lymph node metastasis was detected. Nomiya et al.33 reported similar observations. In oral cancers, some investigators reported that an up-regulation of VEGF-C and a high intratumoral LVD value were correlated positively with lymph node metastasis.12, 34 In our investigation, VEGF-C expression was correlated significantly with T classification but did not influence other clinicopathologic parameters (age, sex, tumor size, histologic grade, lymph node involvement, pathologic stage). In addition, we identified no correlation between LVD or MVD and VEGF-C, suggesting that VEGF-C is not the primary or the only vascular stimulator involved in the progression of oral cancer, particularly as reported by other investigators.35

Generally, high LVD and MVD values were correlated significantly with tumor site, lymph node involvement, mode of invasion, and survival rate in the current study. In contrast, other investigators reported that a low LVD value in tumor cells was associated significantly with the presence of lymphatic invasion and lymph node metastasis.18 In this study, higher LVD and MVD values were correlated significantly with lymph node metastasis and lower survival rates.

Therefore, patients who have an elevated LVD value may have an increased risk for lymph node metastasis, supporting the idea that LVD is responsible for the predominant lymphatic spread in oral cancer. A high LVD value may identify patients who are more susceptible to lymphogenous spread. Thus, preoperative biopsies followed by additional staining for D2-40 and CD105 may be important prognostic indicators for disease progression and may be crucial for deciding on therapeutic strategies for patients with oral cancer.


We thank Emeritus Professors M. Kitano and H. Hiai for their critical reading of the article and Mrs. F. Kataoka for her technical help in the histologic examinations.