Overexpression of podoplanin in oral cancer and its association with poor clinical outcome

Authors

  • Ping Yuan MD,

    1. Molecular Biology Laboratory, Department of Thoracic/Head and Neck Medical Oncology, University of Texas M. D. Anderson Cancer Center, Houston, Texas
    2. Department of Pathology, Ruijin Hospital, Shanghai Jiaotong University, Shanghai, People's Republic of China
    Search for more papers by this author
  • Stephane Temam MD,

    1. Molecular Biology Laboratory, Department of Thoracic/Head and Neck Medical Oncology, University of Texas M. D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Adel El-Naggar MD,

    1. Department of Pathology, University of Texas M. D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Xian Zhou PhD,

    1. Department of Biostatistics and Applied Mathematics, University of Texas M. D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Diane D. Liu MS,

    1. Department of Biostatistics and Applied Mathematics, University of Texas M. D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • J. Jack Lee PhD,

    1. Department of Biostatistics and Applied Mathematics, University of Texas M. D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Li Mao MD

    Corresponding author
    1. Molecular Biology Laboratory, Department of Thoracic/Head and Neck Medical Oncology, University of Texas M. D. Anderson Cancer Center, Houston, Texas
    2. Cancer Biology Program, University of Texas Graduate School of Biomedical Sciences at Houston, Houston, Texas
    • Department of Thoracic/Head and Neck Medical Oncology, Unit 432, University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Blvd., Houston, TX 77030
    Search for more papers by this author
    • Fax: (713) 792-1220


Abstract

BACKGROUND.

Podoplanin is a mucin-like glycoprotein that is important in lymphangiogenesis but not blood vessel formation. Recent studies suggested a potential role of podoplanin in certain tumor cells. The purpose of the current study was to determine the role of podoplanin in head and neck squamous cell carcinoma (HNSCC).

METHODS.

Podoplanin expression was analyzed in 35 patients with HNSCC including 16 oral tumors and 19 hypopharyngeal tumors by immunohistochemical analysis and the association between the podoplanin expression status and patients' clinical and pathologic characteristics was evaluated. An independent set of 60 patients with oral tongue cancer was then analyzed for associations between the podoplanin expression status and patients' clinical and pathologic characteristics, including survivals.

RESULTS.

Podoplanin was not expressed in normal oral epithelial cells but was detected in some hyperplastic and dysplastic lesions. High podoplanin expression was found in 20 (57%) of the 35 tumors and was more frequent in tumors with lymph node metastasis, particularly for tumors in the oral cavity. In the second set of 60 oral tongue cancers, 36 (60%) expressed high levels of podoplanin. Patients whose tumors expressed high levels of podoplanin had a statistically significantly higher rate of lymph node metastasis (P < .0001). Patients with lymph node metastasis and high-level podoplanin showed the shortest disease-specific survival (P = .0004) than other patients.

CONCLUSIONS.

Podoplanin is involved in oral tumorigenesis and may serve as a predictor for lymph node metastasis and poor clinical outcome. Cancer 2006. © 2006 American Cancer Society.

Head and neck squamous cell carcinoma (HNSCC) is 1 of the most common types of cancer, with an annual incidence of more than 500,000 cases worldwide. In the US alone, approximately 28,000 men and 12,000 women are diagnosed with HNSCC each year, representing 3.2% of all newly diagnosed cancers and resulting in 2.1% of cancer-related deaths.1 Patients with HNSCC experience severe disease- and treatment-related morbidity and have only a 50% 5-year survival rate, which has not improved in more than 2 decades.2

HNSCC is a complex disease with various origins, including the oral cavity, pharynx, and larynx. The anatomic location of HNSCC is important for its clinical classification and for treatment selection because tumors arising from different sites have distinct clinical presentations and treatment outcomes. Although the current TNM staging system is used routinely, its value in predicting clinical outcomes remains modest, particularly among tumors of the same TNM stage.3 Bundgaard et al.,4 for example, found that 25% of T1 oral squamous cell carcinomas behaved aggressively and had a poor clinical prognosis. Therefore, it is important to further understand the biology of HNSCC development and to identify biologic markers that may be able to augment the clinical staging system.

Lymph node metastasis is the single most adverse independent prognostic factor in patients with HNSCC.5 However, cervical lymph node metastasis cannot always be predicted from the size and extent of invasion of the primary tumors. Some patients with small primary tumors have cervical lymph node metastasis, whereas some with large tumors have no regional lymph node invasion. It is believed that the underlying molecular features of the tumors play an essential role in determining the aggressiveness of the tumors.

Podoplanin is a mucin-like transmembrane glycoprotein that is highly and specifically expressed in lymphatic endothelial cells, but not blood endothelium.6 It has been shown that podoplanin deficiency disrupts normal lymphatic vasculature formation and causes lymphedema in a podoplanin knockout animal model.7 Recent studies have also shown that podoplanin may be expressed in certain tumor cells, including squamous cell carcinoma cells,8–10 raising a possibility that podoplanin may have biologic functions in tumor cells.

In this study we used monoclonal antibody D2-40, which specifically recognizes podoplanin, to determine the expression of podoplanin in patients with primary HNSCC and potential associations between the expression and clinical and pathologic characteristics. We then tested an independent set of patients with oral tongue cancers to confirm and further elucidate the role of podoplanin in oral cancers.

MATERIALS AND METHODS

Patients

Two patient populations were used in this study. The first included 35 patients with Stage III or IV HNSCC who underwent surgery at the Institute Gustave-Roussy (Villejuif, France) between January 2000 and September 2001. Thirty patients (86%) underwent postoperative radiotherapy or chemoradiotherapy. The clinical characteristics of the patient population are listed in Table 1; 16 patients had oral cavity tumors and 19 had hypopharyngeal tumors. The clinical follow-up data were not available for the population.

Table 1. Clinical Characteristics of the Two Patient Groups
CharacteristicFirst set N = 35 (%)Second set N = 60 (%)
  • *

    T stage in 2 patients could not be determined.

Sex
Female9 (26)22 (37)
Male26 (74)38 (63)
Tumor site
 Oral cavity16 (46)60 (100)
 Hypopharynx19 (54)0
Differentiation
 Well15 (43)15 (25)
 Moderate12 (34)35 (58)
 Poor8 (23)10 (17)
T stage*
 T1 or T2040 (69)
 T3 or T435 (100)18 (31)
N stage
 pN010 (29)32 (53)
 pN1-225 (23)28 (47)
Pathologic stage
 I or II0 (0)24 (40)
 III or IV35 (100)36 (60)
Podoplanin expression
 High20 (57)36 (60)
 Low15 (43)24 (40)

Because the results from the initial set of patients showed more promising data for patients with oral cancers, we focused our second set of patients on 60 oral tongue cancers available for the study. These patients had undergone surgery for Stage I-IV oral tongue cancer at the University of Texas M. D. Anderson Cancer Center (Houston, TX) between 1991 and 1994. Twenty-eight (47%) of these patients had lymph node metastasis (pathologic staging) and had undergone postoperative radiotherapy. Survival data were available for all 60 patients, with a median follow-up after surgery of 6.48 years. The institutional review boards of both institutions approved the study.

Immunohistochemical Analysis

Tissue sections (4 μM) from paraffin-embedded, formalin-fixed tissue blocks of excised tumors were mounted on positively charged glass slides. An immunohistochemical analysis was performed using the avidin-biotin peroxidase complex technique. In brief, the tissue sections were deparaffinized in xylene and rehydrated in graded alcohols. Slides were steamed for antigen retrieval with 10 mM citrate buffer, pH 6.0 (Dako Cytomation, Carpinteria, CA), for 20 minutes. After cooling for 45 minutes, the slides were immersed in methanol containing 3% hydrogen peroxide for 10 minutes to block the endogenous peroxidase activity; this was followed by incubation in 10% horse serum for 30 minutes at ambient temperature to reduce nonspecific binding. The slides were then incubated with the monoclonal antibody D2-40 at a 1:100 dilution at 4°C overnight, followed by processes for standard avidin-biotin immunostaining, according to the manufacturer's protocol (Vectastatin Elite ABC kit, Vector Laboratories, Burlingame, CA). Antibody detection was performed with diaminobenzidine chromogen substrate solution (Vector Laboratories). The slides were counterstained with Mayer hematoxylin (Dako Cytomation). Adjacent normal-appearing lymphatic endothelial cells within the sections served as positive internal controls.

For scoring, representative areas of each tissue section were selected and evaluated independently by 2 investigators (P.Y. and S.T.), who were blinded to the clinical information pertaining to the subjects. Cytoplasm and/or membrane immunoreactivity were considered to indicate podoplanin expression. Quantity scores of 0 to 5 were given if 0%, 1% to 10%, 11% to 30%, 31% to 50%, 51% to 80%, and 81% to 100% of the tumor cells were positive, respectively. The staining intensity was rated on a scale of 0 to 3, with 0 = negative, 1 = weak, 2 = moderate, and 3 = strong. The raw data were then converted to a German Immunoreactive Score (IRS) by multiplying the quantity and staining intensity scores.11 Theoretically, the scores could range from 0 to 15. An IRS score of 8 or higher was considered high reactivity, 4 to 7 moderate, and 0 to 3 weak. Each investigator categorized the cases as high, moderate, weak, or negatively reactive on the basis of IRS. Interobserver differences were minimal. The consensus opinions were used to assign final IRS scores to the disputed cases before data analysis.

Statistical Analysis

Differences in continuous variables grouped by patients' clinical and pathologic characteristics were assessed using the Wilcoxon rank sum test. A chi-square test or Fisher exact test was used to assess the associations among categorical data. Clinical and pathologic characteristics were analyzed for their association with survival using Cox proportional hazards models. Estimates of survival curves were calculated according to the Kaplan–Meier product-limit method. Survival times among patients with different characteristics were compared by means of the log rank test. Predictive variables with P-values of less than 0.10 for the univariate Cox proportional hazards model were included in a multivariate model and a significance level of 0.10 was used to determine the independent factors. All computations were carried out with SAS (Cary, NC) and S-Plus 2000 (Cambridge, MA) software on a Dell computer using the Windows NT operating system. A P-value ≤0.05 was considered statistically significant.

RESULTS

Expression Patterns of Podoplanin in HNSCC

As expected, podoplanin was highly expressed in the endothelial cells of lymphatic vessels, but was not detectable in the endothelial cells of blood vessels (Fig. 1). Podoplanin was expressed mainly in the cytoplasm and membrane. In histological normal squamous epithelium adjacent to the tumors, podoplanin expression was not detectable (Fig. 1A-C) or was extremely low in basal cells. However, in some of the hyperplastic and dysplastic epithelial areas adjacent to the tumors, podoplanin was highly expressed in the basal cell layers (Fig. 1D-F).

Figure 1.

Podoplanin expression in epithelium adjacent to head and neck squamous cell carcinoma (HNSCC). (A-F) Podoplanin was expressed in lymphatic endothelial cell. Podoplanin expression was not detectable in oral epithelium with (A-C) normal histologic characteristics, but was observed in the basal layer of some (D) hyperplastic and (E,F) dysplastic lesions adjacent to tumor.

In primary HNSCC tumors, podoplanin expression was generally heterogeneous and was displayed in 2 patterns: diffuse expression in most living tumor cells (Fig. 2A,B) and focal expression at the proliferating periphery of the tumor cell nests with no expression in the central areas (Fig. 2C). In the latter cases the central areas often contained more differentiated cells, mimicking the pattern seen in the hyperplastic and dysplastic epithelium (Fig. 1D-F). Some of the tumors exhibited weak or no podoplanin expression (Fig. 2D). We observed small tumor nests in a regional lymph node metastasis (Fig. 2E) and tumor cells inside lymphatic vessels with positive podoplanin expression (Fig. 2F).

Figure 2.

Podoplanin expression in head and neck squamous cell carcinoma (HNSCC). (A,B) Diffuse expression pattern in tumors. (C) Focal expression pattern at the growth edge of tumor nests. (D) Weak expression was observed in this tumor. (E) Small tumor nest in a lymph node with podoplanin expression. (F) Podoplanin-positive tumor nest inside a lymphatic vessel.

Podoplanin Expression in Tumors and Its Association with Clinical and Pathologic Characteristics in Patients with HNSCC

In the first set of 35 HNSCC tumors, 2 (6%) lacked podoplanin expression, 13 (37%) had weak or moderate expression (IRS scores 0–7), and 20 (57%) had high expression (scores 8–15). For the purpose of statistical analysis, tumors with scores equal to or higher than 8 were considered to have high podoplanin expression, whereas those with scores lower than 8 were considered to have low podoplanin expression.

Because 71% of the patients had lymph node metastasis, particularly the patients with hypopharyngeal tumors (17 of 19, 89%), statistical evaluation of the relation between podoplanin expression and lymph node metastasis was difficult in the relatively small sample size. Nevertheless, patients whose tumors expressed high levels of podoplanin had a higher frequency of lymph node metastasis than did those whose tumors expressed low levels of podoplanin. Overall, 17 (85%) of the 20 patients whose tumors had high podoplanin expression had lymph node metastasis compared with 8 (53%) of the 15 patients whose tumors had low podoplanin expression, with marginal statistical significance (P = .06). Among the 16 patients with oral cancers, 6 (75%) of the 8 patients with high podoplanin expression had lymph node metastasis compared with only 2 (25%) of the 8 patients with low podoplanin expression, but the difference was not statistically significant (P = .13), probably because of the smaller sample size. No statistically significant associations were observed between podoplanin expression and other clinical and pathologic characteristics (data not shown).

Podoplanin Expression and Lymph Node Metastasis in Patients with Oral Tongue Cancer

To further determine the role of podoplanin in oral cancers, we analyzed an independent set of patients with oral tongue cancer. Two (3%) of the 60 tumors had no podoplanin expression, 22 (37%) had weak or moderate expression, and 36 (60%) had high expression on the basis of the established scoring and classification criteria in our first experiment described above. Among the 28 patients with lymph node metastasis, 25 (89%) had high levels of podoplanin in their primary tumors compared with 11 (34%) of the 32 who had no detectable nodal metastasis (P < .0001, Table 2). In addition, the high podoplanin expression was associated with higher pathologic stage (P = .007, Table 2). No association was found with other clinical and pathologic characteristics (Table 2).

Table 2. Correlations between Podoplanin and Patient Characteristics
 PodoplaninP
Low (n = 24)High (n = 36)
  1. Std, standard deviation.

Age.70
Mean ± std57.8 ± 12.457.36 ± 11.1
Median (min, max)60.9 (38.1, 78.5)56.7 (38.0, 77.2)
Sex (%).51
 Female10 (45.5)12 (54.5)
 Male14 (36.8)24 (63.2)
Grade(%).12
 Well7 (46.7)8 (53.3)
 Moderate16 (45.7)19 (54.3)
 Poor1 (10)9 (90)
T stage(%).94
 T1+T216 (40)24 (60)
 T3+T47 (38.9)11 (61.1)
N stage(%)<.0001
 N021 (65.6)11 (34.4)
 N1+N23 (10.7)25 (89.3)
Pathologic stage (%).007
 IA+IB15 (62.5)9 (37.5)
 IIA+IIB9 (25)27 (75)

Podoplanin Expression and Survival Durations in Patients with Oral Tongue Cancer

To determine whether podoplanin expression in tumors can be used to predict patients' clinical outcome, we compared patients' podoplanin expression status and overall and cancer-specific survival durations. Although patients whose tumors expressed high levels of podoplanin had a shorter overall survival than did those whose tumors expressed low levels of podoplanin, the difference was not statistically significant (P = .11). But the difference in disease-specific survival was statistically significant (P = .001, log rank test) (Fig. 3A).

Figure 3.

Probability of disease-specific survivals in patients with primary oral tongue cancers. The Kaplan-Meier method was used to determine the survival probability and the log rank test was used to compare the survival curves between groups. (A) Probability of disease-specific survival between patients whose tumors expressed high levels of podoplanin and those with low levels. (B) Probability of disease-specific survival among Group 1 (patients without lymph node metastasis and with low levels of podoplanin in primary tumors), Group 2 (patients with either lymph node metastasis or high levels of podoplanin in primary tumors), and Group 3 (patients with both lymph node metastasis and high levels of podoplanin in primary tumors). E, number of patients with disease events; N, total number of patients in the group.

We then performed a Cox proportional hazards regression analysis to determine whether the effect of podoplanin expression on disease-specific survival is dependent on other known risk factors. In the univariate analysis, several factors were statistically significantly associated with survival, including T stage, N stage, pathologic stage, and podoplanin expression (Table 3). In the subsequent multivariate analysis, both N stage and podoplanin show a trend toward a worse disease-specific survival (P = .05 and P = .09, respectively) (Table 4). In a survival analysis, we compared patients with nodal metastasis and high-level tumor podoplanin expression, patients with either nodal metastasis or high-level tumor podoplanin, and patients without any of the 2 events. Patients with both nodal metastasis and high-level tumor podoplanin had the shortest disease-specific survival compared with 2 other groups (P = .0004) (Fig. 3B).

Table 3. Univariate Cox Proportional Hazards Regression Models in Estimating Disease-Specific Survival
CharacteristicsParameter estimateStd. errorPHazard ratio
  1. Std, standard deviation.

Age0.020.021.381.02
Gender (male vs. female)0.290.501.571.33
Grade
 Moderate vs. well0.630.77.421.87
 Poor vs. well1.550.82.064.73
T stage (T3+T4 vs. T1+T2)1.010.50.042.75
N stage (N1+N2 vs. N0)2.000.63.0027.39
Pathologic stage (III+IV vs. I+II)1.460.64.024.30
Podoplanin (high vs. low)2.571.03.0113.06
Table 4. Multivariate Cox Proportional Hazards Regression Models in Estimating Disease-Specific Survival
CharacteristicsParameter estimateStd. errorPHazard ratio
  1. Std, standard deviation.

N stage (N1+N2 vs. N0)1.320.67.053.73
Podoplanin (high vs. low)1.831.10.096.25

DISCUSSION

The monoclonal antibody D2-40 was initially developed to recognize the M2A antigen, which is an oncofetal glycoprotein expressed by testicular germ cell neoplasm.12 Because M2A is found in lymphatic endothelial cells but not blood endothelial cells, D2-40 has been used to detect lymphatic vessels in recent studies.13, 14 The specific staining of lymphatic vessels in the tissue sections from our study is consistent with that found in previously published studies. What surprised us was the strong D2-40 staining in oral and hypopharyngeal tumor cells. Our initial goal was to use D2-40 to stain lymphatic vessels and determine whether the number of lymphatic vessels in and around tumors might predict regional lymph node metastasis. However, the extensive staining made counting the lymphatic vessels inside the tumors impossible. Because many of our tumor sections did not have intact surrounding regions, it was also difficult to assess the association between the number of lymphatic vessels and lymph node metastasis and the association between the numbers of lymphatic vessels and tumor D2-40 staining patterns.

A recent study, using multiple approaches, found that the D2-40 antibody specifically recognizes human podoplanin, which has biochemical characteristics similar to the M2A antigen.8 The assignment of M2A to podoplanin is important, not only making D2-40 a valuable reagent for studying podoplanin but also allowing better data interpretation to determine the biologic roles of podoplanin in physiology and possibly tumorigenesis. Although the biologic functions of podoplanin are not fully understood, overexpression of the protein can promote the formation of elongated cell extensions and increase adhesion, migration, and tube formation of vascular endothelial cells,7 suggesting a role for podoplanin in cytoskeletal reorganization. In a mouse model, the podoplanin homolog PA2.26 was found to be up-regulated during epidermal carcinogenesis,15 suggesting a role of the class molecules in epithelial tumor progression.

In this study we found that podoplanin was expressed in most of the HNSCC cases and was strongly associated with lymph node metastasis. More important, the high levels of podoplanin expression were associated with decreased patient survival, particularly disease-specific survival, in patients with oral cancers. These results are consistent with the findings from cell culture and animal experiments, supporting the importance of podoplanin in oral tumorigenesis.

One unique feature of our study is the 2-stage experimental design, which incorporated testing and validation steps. In the first stage we observed an association between high levels of podoplanin in tumors and lymph node metastasis, which was more prominent in patients with oral cancers. On the basis of this finding, we proposed a specific hypothesis and tested it using an independent set of patients with oral tongue cancers. Because we used the same techniques, reagents, and data interpretation criteria in the second stage of our study, the finding that high levels of podoplanin expression are associated with lymph node metastasis and predict poor clinical outcomes in patients with oral tongue cancers is more credible. These results suggest that podoplanin may play a role in promoting the spread of tumor cells through lymphatic vessels. Indeed, we observed that tumor cells invaded the regional lymph nodes expressing high levels of podoplanin. Podoplanin-expressing tumor cells were also observed inside lymphatic vessels. Taken together, our findings suggest that podoplanin may be useful as a biomarker for the molecular classification of oral cancers. Nevertheless, further studies with larger sample sizes are needed to validate our findings. Furthermore, a causal relation between podoplanin and lymph node metastasis needs to be determined in additional experiments.

The observation that podoplanin is highly expressed in some hyperplastic and dysplastic lesions adjacent to the primary tumors indicates that the overexpression occurs early in head and neck tumorigenesis. More studies are needed to determine the role of podoplanin in head and neck tumor initiation and progression and to determine whether overexpression of podoplanin in head and neck premalignant lesions such as oral leukoplakia may serve as a biomarker to predict the development of invasive cancer. We have previously demonstrated that multiple oral premalignant lesions may be the result of the spreading of index lesions in some patients.16 It would also be interesting to investigate whether podoplanin is involved in such process.

Ancillary