EMMPRIN Expression in Oral Squamous Cell Carcinomas: Correlation with Tumor Proliferation and Patient Survival

The aim of our study was to explore the clinicopathological and prognostic significance of extracellular matrix metalloproteinase inducer (EMMPRIN) expression in oral squamous cell carcinomas (OSCC), and its relation with the proliferative tumor status of OSCC. We examined EMMPRIN and Ki-67 proteins expression by immunohistochemistry in 74 cases with OSCC. Statistical analysis was conducted to examine their clinicopathological and prognostic significance in OSCC. EMMPRIN membrane expression was observed in all cases, with both membrane and cytoplasmic tumor expression in 61 cases (82.4%). EMMPRIN overexpression was observed in 56 cases (75.7%). Moderately or poorly differentiated tumors showed EMMPRIN overexpression more frequently than well-differentiated tumors (P = 0.002). Overexpression of EMMPRIN was correlated with high Ki-67 expression (P = 0.004). In the multivariate analysis, EMMPRIN overexpression reveals an adverse independent prognostic value for cancer-specific survival (CSS) (P = 0.034). Our results reveal that EMMPRIN protein is overexpressed in more than two-thirds of OSCC cases, especially in high proliferative and less differentiated tumors. The independent value of EMMPRIN overexpression in CSS suggests that this protein could be used as an important biological prognostic marker for patients with OSCC. Moreover, the high expression of EMMPRIN makes it a possible therapeutic target in OSCC patients.


Introduction
Oral cancer remains a major public health problem with almost 300,000 new cases worldwide [1,2]. New insights in cancer diagnosis and therapy have not changed significantly the survival rate for oral cancer (around 50%) during the last decades [1].
Oral tumorigenesis is a multistep process caused by accumulation of multiple genetic and epigenetic alterations [3]. The comprehension of the molecular pathways involved in this process may originate special biological markers able to differentiate tumors with a more or less aggressive behavior. These markers may contribute to identify and stratify patients with greater precision to the most appropriate treatment plan.
Moreover, these molecules could become molecular therapeutic targets.
Previous reports have shown that EMMPRIN expression is associated with a high tumor aggressive behavior and with poor prognosis in several tumors [10][11][12][13][14][15][16][17][18][19]. However, in OSCC the prognostic significance of EMMPRIN is poorly studied. Moreover, the relation of this glycoprotein with the proliferative tumor capacity in patients with OSCC has not been reported.
We aimed in this study to evaluate the expression of EMMPRIN in patients with OSCC and investigate the association of this glycoprotein with clinicopathological, tumor proliferation, and prognosis variables.  [20]. For all tumors, 3 m sections were cut and stained with haematoxylin-eosin (HE) to confirm the initial diagnosis. Tumor grade was reclassified following the WHO classification (2005) into well-differentiated (G1), moderately differentiated (G2), and poorly differentiated (G3) OSCC [21]. Inspection for possible presence of tumor lymphatic invasion and perineural permeation reported as present or absent was performed on each sample.

Tissue Microarray (TMA) Construction.
Immunohistochemistry was performed on tumor tissues using TMA technology designed and constructed according to rules previously described [22]. Briefly, representative tumor areas were selected on haematoxylin and eosin-stained sections and marked on paraffin blocks, avoiding necrosis and keratin areas. Three cylindrical tissue cores (2 mm in diameter) were obtained from each selected specimen and transferred to a recipient paraffin block, using a microarray instrument (TMA Builder, Histopathology Ltd., Hungary). From each TMA block, 3 m sections were cut and processed for immunohistochemistry.

2.3.
Immunohistochemistry. TMA slides were deparaffinised in xylene, dehydrated in an ethanol series, and rinsed in distilled water. Epitope retrieval treatment was performed using 0.01 M citrate buffer (pH 6.0) for CD147 and 0.01 M trisethylenediaminetetraacetic acid (EDTA) buffer (pH 9.0) for Ki-67 at high temperature (98 ∘ C water bath during 30 minutes). After blocking endogenous peroxidase with methanol containing 0.3% hydrogen peroxide (H 2 O 2 ) for 5 min, sections were incubated with a blocking solution made of 0.4% casein in trisbuffered saline (TBS) to reduce nonspecific binding. TMA slides were incubated with the primary monoclonal antibody (anti-CD147, clone AB1843, Novocastra, Newcastle upon Tyne, UK, diluted at 1 : 30; and anti-Ki-67, clone MIB1, Dako, Glostrup, Denmark, diluted at 1 : 10) during 60 minutes at room temperature. Then the slides were washed in TBS, followed by incubation with standard peroxidase-labelled dextran polymer for visualization with diaminobenzidine as chromogen (NovoLink Polymer Detection System, Novocastra, Leica Biosystems Newcastle Ltd.), according to the manufacturer's instructions. TMA tissue sections were lightly counterstained with Mayer haematoxylin for 2 min and cover-slipped. Positive (skin and oral mucosa) and negative (omission of primary antibody) controls were used in each staining run.

Evaluation of Immunohistochemical Expression.
All samples were evaluated by two authors blinded to clinicopathological characteristics. The discordant cases were reviewed under a multihead microscope to achieve a consensus. We used the higher score out of at least 2 of the 3 cores examined per case. EMMPRIN staining was evaluated on the basis of extent and intensity immunolabeling of tumor cells. The intensity of staining was scored as 0 (absent), 1 (weak), 2 (moderate), and 3 (strong). The extent of membrane tumor cells staining was semiquantitatively evaluated as 0 (no labelling or labelling in <10% of tumor cells); 1 (labelling in 10% to 24% of tumor cells); 2 (labelling in 25% to 49% of tumor cells); 3 (labelling in 50% to 74% of tumor cells); and 4 (labelling in 75% or more of tumor cells). The sum of the intensity and extent scores was used as the final score (0-7).  Tissues having a final score of 0-1 were considered negatives. Final scores of 2-3, 4-5, and 6-7 were considered 1+, 2+, and 3+, respectively. For data analysis, score 3+ was defined as EMMPRIN overexpression [12]. For Ki-67 evaluation, we considered the percentage of nuclear staining for scoring proliferative status. We classified tumors into two groups: low proliferative tumor (labelling from 0 to 49% of tumor cells) and high proliferative tumor (labelling in 50% or more of tumor cells) [23].

Statistical Analysis.
Statistical analysis was carried out using IBM SPSS Statistics version 21.0 software (IBM Corporation, NY, US). The associations between categorical variables were evaluated by chi-square tests. Correlation between EMMPRIN and Ki-67 was measured by Spearman's correlation coefficient. Cancer-specific survival (CSS) was defined as the time interval (months) between primary treatment and death from oral cancer or last follow-up. Recurrence-free survival (RFS) was defined as the time interval (months) between primary treatment and the first recurrence (whether local, regional, or distant). The Kaplan-Meier method was used to plot survival curves and their prognostic effect was tested using the log-rank test. Variables with significant effects in the univariate analyses were entered into Cox proportional hazards model to investigate the independent effects of these variables. Differences were considered statistically significant at < 0.05.

EMMPRIN Expression.
Immunohistochemistry was performed in 74 human OSCC tissues to evaluate the extent and patterns of EMMPRIN protein expression. All cases presented membrane staining for EMMPRIN on tumor cells. Additionally, in 61 cases (82.4%), cytoplasmic expression was also observed. On the basis of EMMPRIN immunoexpression, cases were classified as 1+ in 2 (2.7%), 2+ in 16 (21.6%), and 3+ in 56 (75.7%) cases (Figures 1(A) and  1(B)). Staining of this protein was detected predominantly at the periphery of the tumor islands (45; 60.8%) or present homogenously within the tumor islands (29; 39.2%) (Figures  1(A) and 1(B)). We observed also that EMMPRIN expression was seen in peritumoral fibroblasts in 65 (87.8%) cases (Figures 1(A) and 1(B)). Fibroblasts were identified by an experienced pathologist based on its histomorphological features. In cases in which difficulties existed in fibroblast identification, we recurred to coloration for vimentin and smooth muscle actin. Apparently, normal mucosa adjacent to primary tumor presented a strong EMMPRIN staining in basal and suprabasal epithelial layers.
We compared EMMPRIN expression in OSCC tissue samples with patient clinicopathological variables. A positive association of EMMPRIN expression with histological grade was noted where G2/G3 tumors presented EMMPRIN overexpression more often than G1 tumors ( = 0.002) ( Table 1). Although no significant association was found between EMMPRIN overexpression and clinical stage, when we divided clinical stage into initial stage (I/II) tumors and advanced stage tumors (III/IV), we observed a significant association between EMMPRIN overexpression and advanced tumor stages ( = 0.015). No other significant relationship between EMMPRIN expression and the listed clinicopathological parameters was found. Moreover, we did not find significant relationship between EMMPRIN distribution pattern (homogeneously or peripherically), EMM-PRIN fibroblast staining, and the listed clinicopathological parameters. On a univariate analysis using the Kaplan-Meier method and log-rank test, we measured the influence of the clinicalpathological and immunoexpression variables on the survival of patients with OSCC. EMMPRIN overexpression was statistically associated with a worse CSS ( = 0.011) ( Figure 2; Table 2). Among the clinicopathological characteristics, tumor size ( < 0.001), status ( = 0.003), tumor stage ( < 0.001), treatment modality ( = 0.05), and histological grade ( = 0.037) were also statistically associated with a worse CSS ( Table 2). As also described in Table 2, we observed a significant association between RFS and gender ( = 0.013), tumor size ( = 0.009), status ( = 0.006), tumor stage ( = 0.009), margin status ( = 0.003), and perineural permeation ( = 0.041).

EMMPRIN and
In the multivariate analyses using Cox regression method, we found an association of EMMPRIN overexpression with poor survival ( = 0.034; Table 3), thus revealing EMMPRIN overexpression as an adverse independent prognostic factor for CSS in OSCC. In RFS, gender ( = 0.030) and margin status ( = 0.019) reveal an independent prognostic value for these tumors (Table 4).

Discussion
Recent studies have reported the biological and clinical role of EMMPRIN receptor in several cancers in the last decades [4,5,24]. However, the influence of this receptor in OSCC is poorly understood. In the present study, we aimed to evaluate the expression of EMMPRIN protein in OSCC and to analyse the correlation of this receptor with clinicopathological characteristics, tumor proliferation, and patient's outcome.
Our study showed that EMMPRIN protein was present in all OSCC cases and overexpressed in more than two-thirds of the cases. This result is in accordance with the notably high expression of this glycoprotein in squamous cell carcinomas of head and neck region reported by Riethdorf et al. [5]. By analyzing EMMPRIN expression in multitumor TMAs, they observed expression in more than 95% of squamous cell carcinoma of oral cavity and 100% in squamous cell carcinomas of salivary glands. Lower expression values have been reported by Gou et al. [25] in laryngeal carcinomas (87.5%), Zhu et al. [17] in esophageal squamous cell carcinomas (85%), and Huang et al. [11] in tongue squamous cell carcinomas (67%). They also observed a significantly higher expression on tumor cells than in the noncancerous epithelium. Vigneswaran et al. [26] found a strong EMMPRIN expression in more than 90% of tumor cells in carcinoma in situ and early-invasive OSCC and also a significant higher expression compared with normal oral mucosa. The authors also found an increasing expression of this marker in oral leukoplakias gradually correlated with the degree of dysplasia, suggesting that EMMPRIN overexpression occurs at an early step of oral carcinogenesis and contributes to oral tumorigenesis. These data highlight the potential important role of EMMPRIN in OSCC.
We observed EMMPRIN expression in tumor cell membrane and also in the cytoplasm of some cases, in concordance with other studies [12,14,19]. Although our aim was   to analyse the expression of EMMPRIN in tumour cells, we detected the presence of this receptor in peritumoral fibroblasts as described by Vigneswaran et al. [26]. Furthermore, EMMPRIN expression showed a predominantly peripheric/basal distribution pattern in the tumor islands in most of our cases. This was reported in other works suggesting a more frequent distribution of this receptor in tumor cells with a more proliferative phenotype [11,26]. In order to assess the relationship of EMMPRIN with proliferative activity, we evaluated the expression of Ki-67 in these tumors and found a significant correlation between these two proteins. To our knowledge, this correlation has not been reported in OSCC, although Yang et al. [14] described a positive association between EMMPRIN expression and Ki-67 index labelling and also with tumor size in adenoid cystic carcinomas. Similar results were reported by Zheng et al. [27] showing a positive correlation between the two markers in gastric carcinomas. These results are in line with ours, suggesting that EMMPRIN might be relevant for the tumor proliferation and tumor growth of OSCC. Furthermore, knockdown of EMMPRIN in head and neck carcinomas decreased cellular proliferation and tumor growth in vitro and in vivo analyses [28][29][30]. Mechanisms involved in tumor proliferation via EMMPRIN are poorly understood, but some authors have described the role of this receptor in association with cyclophilin A in the activation of ERK1/2 and p38 pathways [31]. We found that EMMPRIN expression was significantly associated with histological grade. Moderately or poorly differentiated tumors showed more EMMPRIN overexpression than well-differentiated tumors. Zhu et al. [17] observed the same positive association in 86 esophageal squamous cell carcinomas. Clinical stage and tumor size have been positively related with EMMPRIN expression in several cancers including head and neck cancers [11,12,25]. We observed that EMMPRIN overexpression was more frequently found in patients with advanced clinical stage (III/IV), emphasizing the biological significance of this marker to tumor growth and progression of OSCC. We did not find any other significant relation with other clinicopathological parameters although 8 BioMed Research International association with nodal metastasis was observed by others [32].
The influence of EMMPRIN expression on patient's survival has been reported in glioblastomas, seminomas, and other cancers including tongue, salivary gland, esophageal, ovary, colorectal, breast, bladder, and lung cancers [11-15, 17, 19, 33-36]. In our univariate analysis, we found that cases with EMMPRIN overexpression were associated with a lower CSS ( = 0.011) additionally with other clinical variables such as TNM and clinical stage. Nevertheless, in the multivariate analysis for CSS, EMMPRIN protein was the only independent prognostic factor ( = 0.034), revealing the adverse independent impact of EMMPRIN overexpression on the survival of patients with OSCC. To our knowledge, this is the first report of the independent prognostic value of EMMPRIN in a cohort of patients with squamous cell carcinoma of the oral cavity. Previously, Huang et al. [11] described the independent significant influence of this receptor in the overall survival of patients with squamous cell carcinomas of the tongue. This could be an important result suggesting the use of this receptor as a prognostic biomarker in OSCC. Interestingly, some studies report that EMMPRIN might be even a predictive marker of chemoresistance in head and neck carcinomas [8]. The influence of this receptor on patient's prognostic could be related to the multiple biological functions of this protein on tumor cells such as proliferation, migration, invasion, angiogenesis, and dissemination on OSCC [4]. Studies have described the role of EMMPRIN in the stimulation of several metalloproteinases and proangionenic factors from tumor and adjacent stroma cells that could contribute to tumor multistep pathogenesis [27]. It would be interesting to analyse the relationship of EMMPRIN expression and molecules involved in different pathways, such as EGFR, MMP's, and VEGFR's, in a larger sample of OSCC.
The understanding of the different pathways involved in oral tumorigenesis could reveal new candidate target molecules for anticancer drugs. Anti-EGFR targeted therapies are currently available for head and neck cancer but with modest results [37]. New anticancer therapies directed to molecular targets on oral cancer cells are needed and some molecules have been proposed, including EMMPRIN receptor [5,9,11,38]. The high expression of this receptor in OSCC, the cell membrane location, the biological role on tumor growth, invasion, dissemination, and the influence in the patient's prognosis make EMMPRIN a strong candidate for a potential molecular target for monoclonal therapies against this receptor in OSCC. Anti-EMMPRIN molecular therapies showed growth inhibitory effect on head and neck squamous cell carcinoma, alone and in combination with radiotherapy in vitro and in vivo [39]. Sweeny et al. [40] reported a promising extracellular drug conjugate (EDC22), capable of inhibiting HNSCC cell proliferation in vitro and in vivo, with better results than with radiation or cisplatin monotherapy.
In conclusion, our results reveal that EMMPRIN protein is frequently overexpressed in OSCC, especially in high proliferative tumors, suggesting that it might be involved in the growth of these tumors. The independent value of EMMPRIN overexpression in CSS indicates that this protein could be used as an important biological prognostic marker to identify high risk OSCC patients, helping in making the right decision as to the appropriate treatment. Furthermore, the high expression of this receptor could be regarded as potential therapeutic target against OSCC.

Conflict of Interests
There is no potential conflict of interests.