EMMPRIN Expression as a Prognostic Factor in Radiotherapy of Cervical Cancer

Cervical cancer is the second most prevalent cancer in women worldwide and the fifth leading cause of cancer deaths (1). There are >470,000 cases annually and 233,400 deaths (1), 80% of all cases are found in developing countries. In China, there are 137,000 new cases every year (2). Epidemic evidence has shown that the incidence of cervical cancer is closely associated with persistent infection of high-risk human papillomavirus. From human papillomavirus infection to invasive cervical cancer, a series of pathologic changes may occur. Despite recent advances in cancer treatment, radiotherapy as preoperative or postoperative adjuvant or primary treatment remains the most common management for advanced cervical cancer. However, there are ∼30% of recurrent cases after radiotherapy (3), and the primary cause of treatment failure in local cervical cancer is resistance to radiotherapy. An understanding of the molecular determinants of the response of tumors to radiotherapy is important for improving the outcomes of the treatment.

Extracellular matrix metalloproteinase inducer (EMMPRIN; CD147), a highly glycosylated cell surface transmembrane protein, belongs to the immunoglobulin superfamily (4, 5). EMMPRIN has been shown to promote tumor invasion and metastasis via stimulating matrix metalloproteinase synthesis in neighboring fibroblasts (6, 7), to enhance angiogenesis via vascular endothelial growth factor (8), to induce chemoresistant tumor cells via the production of hyaluronan (9), and to confer resistance of cancer cells to anoikis through inhibition of Bim (10). EMMPRIN has also been shown to interact with cyclophilin A (CypA) and to facilitate the malignant cell proliferation via the activation of ERK1/2 and p38 mitogen-activated protein kinases (11). Inhibition of CD147 gene expression via RNA interference could reduce tumor cell invasion and tumorigenicity, and increase chemosensitivity to paclitaxel (12).

Based on the molecular functions of EMMPRIN, we speculated that this protein may play a role in the malignant phenotype of cervical cancer. The goal of the present study was to investigate the expression of EMMPRIN in relation to brachytherapy and survival in patients with cervical cancer treated with preoperative brachytherapy, and evaluated the expression of this protein as a prognostic factor in radiation therapy.

Patient recruitment. Patients were selected according to the following criteria: (a) presentation to the Department of Gynecological Oncology at The Cancer Hospital of Fudan University (Shanghai, China) between 1991 and 2000 with a histologic diagnosis of cervical carcinoma; (b) treatment with preoperative intracavity brachytherapy with 15 Gy/3 fractions/2 weeks to point A in high-dose rate using afterloading ¹⁹²Ir radioactive resource followed by radical hysterectomy with pelvic lymphadenectomy; and (c) available archival cervical cancer samples before and after brachytherapy. Patients were excluded from this study if they had any preoperative chemotherapy. The hospital records, pathology reports, and histology slides of each patient were carefully reviewed. Eighty-two patients who met the entry criteria were identified. Covariables including patients' demographics, clinical, pathologic and staging information, and treatment variables were extracted from patient charts.

With the approval of the Institutional Review Board of Shanghai Medical College, paraffin-embedded tissue samples were obtained from hospital archives. The sample before brachytherapy was obtained by punch biopsy at initial diagnosis and the sample after brachytherapy was obtained from radical hysterectomy. Each cervical primary tumor size was assessed by direct measure or imaging study. All patients were clinically staged according to the International Federation of Gynecology and Obstetrics staging criteria (1994). According to our management guideline on cervical cancer, each patient in this study received preoperative brachytherapy with 15 Gy to point A for local bulky tumor (International Federation of Gynecology and Obstetrics stage Ib2-IIb) and then underwent radical hysterectomy and pelvic lymph node dissection. Due to bulky primary cervical tumor (>4 cm in diameter) or other adverse pathologic findings such as metastasis to lymph nodes, lymph vascular space invasion (LVSI), and deep stromal invasion, all of the 82 patients in the study were subject to adjuvant pelvic radiation at a dose of 45 Gy following radical hysterectomy. Clinical details of patient age, disease stage, histologic type and tumor grade are listed in Table 1.

The sample size of this study was determined according to the formula:

U_α = 0.05, U_2β = 1.645, 2β = 0.10, and based on the results of a pilot test in which P₁ (EMMPRIN expression rate before brachytherapy) and P₂ (EMMPRIN expression rate after brachytherapy) were 90% and 70%, respectively. The sample size needed for statistical analysis of significance of this study was 74 cases. Our study included 82 cases, which was a suitable size for statistical analysis.

Immunohistochemistry. Polyclonal antibody against EMMPRIN, and the positive controls (human breast tissues) were purchased from Beijing Zhongshan Goldenbridge Biotechnology Company Limited (ZA-0455). EnVission+, Peroxidase (ready-to-use) from Gene Tech Biotechnology Company Limited (GK400305) was used as the second antibody. Three-micrometer sections of paraffin-embedded tissue were dewaxed and dehydrated before being boiled with Tris-EDTA buffer (1 mmol/L; pH 9.0) at 95°C for 20 min, then washed with PBS. A two-step immunohistochemistry procedure according to the specification sheet (provided by Gene Tech) was done. Titrated concentrations of the first antibody were prepared, sections were then incubated overnight at 4°C in a moist chamber with the antibody against EMMPRIN (at a dilution of 1:50, according to the results of titrated concentrations) and then with the second antibody for 60 min at 37°C. Finally, sections were treated with 3,3′-diaminobenzidine (Gene Tech). Negative control sections were subjected to the same procedure except that the first antibody was replaced by PBS. No significant staining was observed in the controls. The pattern of staining observed in the tumor samples was predominantly cell membrane and cytoplasm.

Staining results were evaluated by two board-certified pathologists. For each tissue sample, the fraction of immunostained tumor cells was recorded, and the staining intensity was estimated on a four-step scale (negative is no staining, 1+ is weak staining, 2+ is moderate staining, and 3+ is strong staining). Tumors were then initially categorized into four groups according to arbitrarily predefined EMMPRIN scores (0, 1, 2, and 3). The exact criteria for these groups were as follows (13): score 0 (no staining at all); score 1 (1+ staining regardless of positive cell percentages or 2+ staining of ≤30% of cells); score 2 (2+ staining of >30% of cells or 3+ staining of ≤50% of cells); and score 3 (3+ staining of >50% of cells).

Statistical analysis. The χ² test was used to analyze the association between EMMPRIN expression and various clinicopathologic variables. For analysis of follow-up data, the Kaplan-Meier method was used. Survival distributions were compared using log-rank statistics. Disease-specific survival was determined from the date of surgery to the time of the last follow-up or cancer-related death. The joint effects with already recognized prognostically relevant variables were examined by Cox regression analysis using a backward LR method. Tumor reduction rate (higher reduction rate represents a lesion reduction of >30% at the greatest dimension), deep stromal invasion (outer 1/2 invasion of the cervical stroma), LVSI, and lymph node metastatic status were entered into the model to test these covariables for possible prognostic effects with respect to EMMPRIN expression (threshold for statistical significance was P = 0.05). All tests of statistical significance were two-sided. For statistical analyses, we used SPSS software for PC (version 12.0 for Windows).

Characteristics of the patient cohort. A total of 82 patients were eligible for this study; 7 were excluded because of incomplete follow-up data, leaving 75 patients (90%) in the survival analysis. The median follow-up time of the surviving patients was 52 months (range, 3-168 months). Thirty-five patients (46.7%) died during the follow-up period. Thirteen (15.9%) patients presented with stage Ib2 disease, whereas 50 (61.0%) had stage IIa, and 19 (23.2%) had stage IIb. Histologically, 72 (87.8%) patients were classified as squamous carcinoma, 6 (7.3%) as adenocarcinoma, 2 (2.4%) as adenosquamous carcinoma, and 2 (2.4%) as clear cell carcinoma, according to the WHO classification. Twenty-two (26.8%) patients were pathologically confirmed to have positive pelvic lymph nodes metastasis. The median age of the patients was 49.7 years (range, 21-72 years).

Distribution of patients according to EMMPRIN expression. The results of EMMPRIN immunohistochemistry staining before brachytherapy were as follows: 43 patients (52.4%) had a score of 3, 27 patients (32.9%) had a score of 2, 7 patients (8.5%) had a score of 1, and 5 patients (6.1%) had no staining. The negative controls exhibited no staining. The pattern of staining observed in the tumor samples was predominantly cell membrane and cytoplasmic, as shown in Fig. 1A and B. To increase the statistical power of this study, the EMMPRIN expression staining variable was dichotomized, with the cut point set between scores of 2 and 3. Thus, patients with a score of 3 were defined as positive for overexpression of EMMPRIN, whereas patients with scores of 2, 1, and no staining were defined as negative for overexpression of EMMPRIN. The distribution of patients according to this criterion is summarized in Table 1. EMMPRIN overexpression in the primary tumor was associated with pelvic lymph node metastasis status (P = 0.026). The frequency of EMMPRIN overexpression in primary tumors was 72.7% (16 of 22), with 54.5% (12 of 22) of metastasis lymph nodes. No significant difference of EMMPRIN overexpression could be seen between metastatic lymph nodes and corresponding primary tumor. The reduction of primary tumor after brachytherapy correlated adversely with EMMPRIN overexpression (P = 0.008). There was no correlation between the level of EMMPRIN overexpression, disease stage, grade, patient age, and histologic type.

Alteration of EMMPRIN expression in cervical cancers before and after radiotherapy. After intracavitary brachytherapy of 15 Gy at point A, an increased expression of EMMPRIN was observed in 11 (13.4%) patients with cervical cancer. Compared with corresponding cervical tumor specimens obtained by biopsy before brachytherapy, EMMPRIN expression of cervical tumor from postoperative samples was decreased in 62 (75.6%) patients, and was increased or remained the same in 20 (24.4%) patients. As shown in Table 2 and Fig. 1A and B, both the frequency and the staining intensity of EMMPRIN expression in postbrachytherapy cervical tumor was significantly decreased (P = 0.032). Our cohort consisted of 72 squamous, 2 adenosquamous, 2 clear cell, and 6 adenocarcinomas, which were selected from >200 cases according to our trial criteria. We found that altered expression of EMMPRIN had no correlation with clinical stage (P = 0.108) and histopathology (P = 1.000). In addition, a shift of EMMPRIN immunostaining from membrane to cytoplasm after brachytherapy was observed (Fig. 1A and B).

EMMPRIN expression and survival. EMMPRIN expression was examined with respect to tumor-specific survival by Kaplan-Meier and log-rank tests. In univariate analysis (Table 3), LVSI, pelvic lymph node metastasis, and deep stromal invasion were the strong prognostic factors for poor outcome. EMMPRIN expression before and after brachytherapy were not associated with prognosis (Figs. 2 and 3). To further examine the relationship between EMMPRIN expression and prognosis, the patients were classified into two groups: increased expression or decreased expression of EMMPRIN in corresponding samples prebrachytherapy and postbrachytherapy; the former was defined as the increased expression of EMMPRIN or no change, and the latter was defined as the decreased expression of EMMPRIN following brachytherapy. The 20 patients with increased EMMPRIN expression had significantly shorter survival than the patients with decreased EMMPRIN expression (P = 0.0008; Fig. 4A). When stratified by clinical stage and histopathology, altered expression of EMMPRIN was defined as a significant prognostic factor in our cohort of patients. When stratified by tumor stage, for stage Ib2 and IIa, altered expression of EMMPRIN was correlated with patients' survival, log-rank = 9.42, P = 0.0021 (there was only a small number of patients with stage Ib2 disease, and they were therefore combined with patients with stage IIa disease), and for stage IIb, log-rank = 6.10, P = 0.0135 (Table 3; Fig. 4B and C). To preclude the effect of histopathologic heterogeneity on the prognostic role of EMMPRIN, we also stratified the cases by histopathology. We found that for squamous carcinoma, altered expression of EMMPRIN was associated with patients survival, log-rank = 9.95, P = 0.0016 (Table 3; Fig. 4D). Due to the small sample size of nonsquamous carcinomas (nine cases, one case ruled out without follow-up data), analysis of survival by Kaplan-Meier could not be done. In all the nine nonsquamous carcinomas, we found that seven cases with decreased EMMPRIN expression had a survival rate of 85.71% (six of seven), and two cases with increased EMMPRIN expression had a survival rate of 50% (one of two). In multivariate analysis using a Cox regression model and a backward likelihood ratio method (Table 4), the alteration of EMMPRIN expression following brachytherapy was one of the most independent predictors of tumor-specific survival (P = 0.002; hazard ratio, 0.339; 95% confidence interval, 0.172-0.672). Lymph metastasis status was another significant independent predictor of tumor-specific survival (P = 0.044; hazard ratio, 2.053; 95% confidence interval, 1.020-4.133).

In the current study, we observed that pelvic lymph node metastases, which represent the tumor invasion and metastatic capability, were significantly associated with EMMPRIN overexpression. These results suggest that EMMPRIN expression in cervical cancer is associated with tumor invasion and metastasis. Our results are consistent with the previous reports of the roles of EMMPRIN in tumor progression (14), including gliomas (15), hepatoma (16), squamous cell carcinoma (17), and melanoma (18, 19). Furthermore, EMMPRIN was shown to be the most frequently expressed protein in primary tumors and in micrometastatic cells (20), and EMMPRIN score was significantly correlated with nodal stage (14), suggesting a central role in tumor progression and early metastasis. To our knowledge, this is the first study that investigates the expression of EMMPRIN and its clinical significance in radiotherapy outcome in cervical carcinoma. Interpretation of the results is aided by the homogeneous radiotherapy before and after radical surgery received by the patients in this study and the availability of complete long-term follow-up survival data.

EMMPRIN is expressed in neoplastic and some normal epithelial cells, although heterogeneity was observed within and between individual tumor biopsies (18, 21–23). In our study, EMMPRIN expression was associated with lymph node metastasis, but not with other clinicopathologic factors. EMMPRIN expression in breast carcinomas is associated with risk factors such as poor histologic grade, negative hormone status, the mitotic index, and tumor size (13). Higher EMMPRIN immunostaining scores in hepatocellular carcinomas correlate significantly with tumor grading and tumor-node-metastasis stages (24). In gastric carcinoma, EMMPRIN expression was positively correlated with tumor size, depth of invasion, and lymphatic invasion, but not with lymph node metastasis, staging, or differentiation (25). However, EMMPRIN protein expression patterns within esophageal squamous cell carcinoma and dysplastic lesions were not associated with any of these clinicopathologic factors (26). These discrepancies suggest that there are different regulatory mechanisms of EMMPRIN expression in cells of different origin.

Studies have shown that EMMPRIN expression seems to correlate with poor prognosis in breast cancer, esophageal squamous cell carcinoma, and ovarian serous cancers, however, it cannot be viewed as an independent prognostic factor (13, 24–28). In this study, although we did not observe that EMMPRIN expression before or after brachytherapy was associated with prognosis, the decreased expression of EMMPRIN after radiotherapy seemed to be a significant predictor of better tumor-specific survival.

It has been shown that EMMPRIN is associated with tumor infiltration and invasion into vessels via up-regulation of matrix metalloproteinases and vascular endothelial growth factor (8), suggesting that blockade of these molecules may prolong disease recurrence–free survival or overall survival by interfering with tumor infiltration and invasion (29, 30). If the activity or expression of EMMPRIN is compromised after radiotherapy, this may explain, at least in part, a decrease in the overall invasive potential in response to radiation and better survival of patients after brachytherapy. In the present study, patients with increased expression of EMMPRIN after brachytherapy had a poor prognosis, which supports this explanation.

An interesting phenomenon which we noticed in our study was that EMMPRIN staining shifted from the membrane to cytoplasm in most cases after radiotherapy (Fig. 1). We believe that this shift is due to the effects of irradiation on cell membrane and membrane proteins, as radiation is known to induce peptide breakage and degradation, alteration of membrane fluidity, change in protein structure, alteration of activity and function of immunoglobulins, etc. (31), and EMMPRIN is an immunoglobulin. The role and mechanism of the shift of EMMPRIN from the membrane to the cytoplasm remains unknown.

We observed, in this study, that EMMPRIN expression in tumor cells was altered by irradiation therapy. Yet, the mechanism underlying this observation is not known. Global gene analysis of radioresistant sublines and other experiments may provide new insights into the mechanisms underlying this phenomenon. Ogawa et al. (32) showed that amphiregulin and mitogen-activated protein kinase/APK2 were among the up-regulated genes in radioresistant pancreatic cancer cell lines established by fractionated irradiation. It has also been identified that amphiregulin and epidermal growth factor were regulatory effectors able to induce both mRNA and protein expression of EMMPRIN by an epidermal growth factor receptor tyrosine kinase activation (33). The up-regulation of EMMPRIN by epidermal growth factor receptor activation is often observed in some malignancies (34). Cervical cancers with increased EMMPRIN expression may result from radioresistant tumor cells whose expression of EGF and amphiregulin are up-regulated during irradiation.

In this study, we observed a relationship between EMMPRIN expression and radioresistance. This might be attributed to the known function of EMMPRIN, such as stimulating the mitogen-activated protein kinase pathway and conferring resistance to anoikis (10, 35–41). Based on our study, patients with locally advanced invasive cervical cancer may be divided into two types according to the level of EMMPRIN expression after brachytherapy: one type with decreased EMMPRIN expression in which a good prognosis could be expected and in which brachytherapy should be the choice for these patients; the other type with increased EMMPRIN expression after brachytherapy in which a poor prognosis can be predicted. For the latter, targeting EMMPRIN might help improve the treatment.

In summary, we found that decreased expression of EMMPRIN in cervical cancer after brachytherapy seemed to be associated with a better prognosis, and increased expression of EMMPRIN following brachytherapy was associated with a poor outcome and a more metastatic and aggressive malignant phenotype. Our results, which link EMMPRIN expression to tumor progression and treatment response, might aid in the development of tailored combined treatments and disease monitoring schedules for patients with cervical cancer. Therefore, EMMPRIN expression may be a novel prognostic marker for locally advanced invasive cervical cancer patients receiving brachytherapy.