Expression Level of Valosin-Containing Protein (p97) Is Associated with Prognosis of Esophageal Carcinoma

There is remarkable geographic variation in both the incidence and the constituents of esophageal cancer. In Asian countries, where esophageal cancer is common, more than 90% of esophageal cancers are squamous cell carcinomas (1, 2, 3). In Western countries, where the incidence of esophageal cancer is relatively low, the incidence rate of esophageal adenocarcinoma has been increasing in the last few decades, whereas that of esophageal squamous cell carcinoma (ESCC) remains constant (4, 5). As a result, ESCC constitutes 50–60% of all esophageal cancers in Western countries at present (4, 5).

ESCC shows a poor prognosis because of the occurrence of systemic metastasis, mainly via lymphatic vessels (1, 2, 3, 4, 5, 6). In Japan, 9991 patients died from ESCC in 1999, accounting for 3.44% of all cancer deaths (2). Detection of ESCC at an early stage is possible with the use of X-ray and endoscopic examinations (7, 8), but there might be occult micrometastases at the time of surgery even in such cases (9, 10). In this respect, an assessment of metastatic potential is important to establish appropriate therapeutic modalities for ESCC.

Previous studies proposed the prognostic significance of several clinicopathologic factors for ESCC, such as tumor size, age at diagnosis, and primary site (5, 6). Among them, depth of cancer invasion in the esophageal wall and lymph node metastasis are the main factors for ESCC recurrence (5, 6, 11), and these two factors together with distant metastasis have been included in the pathological Tumor-Node-Metastasis (pTNM) staging for ESCC (12). However, the prognosis for ESCC, even in cases of early-stage disease (pT₁), is heterogeneous, and a strategy to establish appropriate therapeutic modalities for each patient has yet to be formulated.

To reinforce the prognostic utility of the pTNM classification, several biological indices, such as aberrant expression or mutation of the tumor suppressor p53, adhesion molecule E-cadherin, and cell-cycle-related molecules such as p27, have been proposed (3, 13, 14). However, the utility of these factors for the prediction of prognosis in ESCC is controversial (15).

Recently we identified the gene encoding valosin-containing protein (VCP; also known as p97) as associated with metastatic potential in a murine osteosarcoma cell line using the mRNA subtraction technique (16). VCP, a member of the superfamily ATPases associated with various cellular activities, is known to be involved in the ubiquitin-dependent proteasome degradation pathway (17), which works in both the up-regulation of cell proliferation and the down-regulation of cell death in human cancer cells (18). These findings suggest that the expression level of VCP could be used to predict the metastasis and prognosis of patients with cancers.

In the present study, we analyzed VCP expression in 156 patients with ESCC and evaluated its correlation with clinicopathologic features and survival. In parallel, we examined expression of the tumor suppressor p53 and the proliferative activity of tumor cells shown by Ki-67 immunohistochemistry and evaluated the correlation between VCP expression and these markers. All of the procedures in the present study were approved by the Ethics Committee, Osaka University Graduate School of Medicine.

For the present study, 156 patients who had undergone surgery for ESCC at the Gastroenterological Surgery Division, Osaka University Hospital during the period from July 1989 to February 2002 were selected. There were 140 males and 16 females (age range, 38–82 years; median, 60 years). An endoscopic examination of esophageal lesions was performed, and a histologic diagnosis of ESCC was made based on biopsy specimens obtained before surgery. Preoperative diagnostic examinations, including esophagography, computed tomography, and ultrasound, were performed for the purpose of clinical staging. All patients underwent curative surgery for ESCC. The types of surgery used were subtotal esophagectomy with lymph node resection in 148 patients, partial esophagectomy in 6, and endoscopic mucosal resection in 2. The resected esophagus was examined macroscopically to determine the location and size of the tumor. Tumors were located in the cervical esophagus in 10, the upper third of the thorax in 7, the middle third of the thorax in 80, the lower third of the thorax in 48, and the abdominal esophagus in 11 patients. The size of the main tumor ranged from 2 to 100 mm (median, 44 mm). Samples obtained from the esophageal lesions were fixed in 10% formalin and processed routinely for paraffin embedding; the number of specimens prepared for histologic examination per case ranged from 6 to 46. Histologic sections cut 4-μm thick were stained with hematoxylin and eosin and by immunoperoxidase procedures (the avidin-biotin complex method). Histologic sections were reviewed by one of the authors (Y. H.) to define the depth and mode (expansive or infiltrative) of cancer invasion in the esophagus, lymph node metastasis, and histologic subtype of ESCC based on the criteria of the Japanese Esophagus Society (19). Tumor stages were defined based on the pTNM classification (12).

After surgery, all patients underwent laboratory examinations such as routine peripheral blood cell counts and measurement of the serum squamous cell carcinoma antigen levels every 1–6 months, and chest roentgenography, ultrasonography of the liver, a computerized tomographic scan of the thorax and abdomen, and an endoscopic examination of the remaining esophagus at intervals of 6–12 months. Pre- and postoperative chemotherapy and radiotherapy were performed in 47 patients at high risk for tumor recurrence, i.e., presence of lymph node metastasis, tumor with a diameter larger than 10 cm, and tumor invasion beyond the muscularis propria. The types of adjuvant therapies were as follows: preoperative chemotherapy in 21 cases, radiotherapy in 1 case, and combined chemo- and radiotherapy in 20 cases; postoperative chemotherapy in 1 case, radiotherapy in 1 case, and combined chemo- and radiotherapy in 2 cases; and combined preoperative chemo- and radiotherapy and postoperative chemotherapy in 1 case. The radiation dose ranged from 40 to 60 Gy. The chemotherapeutic agents used were 5-fluorouracil and cisplatinum. The follow-up period for survivors ranged from 1 to 96 months (median, 52.0 months).

The immunoperoxidase procedure (avidin-biotin complex method) was performed with paraffin-embedded sections; one representative section per case was selected. Briefly, antigen retrieval was performed by heating the sections in a 10 mm citrate buffer for 5 min. Mouse monoclonal anti-VCP (p97) antibody (PROGEN Biotechnik, Heidelberg, Germany), anti-p53 monoclonal antibody (Dako Cytomation A/S, Copenhagen, Denmark), and anti-Ki-67 (clone MIB 1) monoclonal antibody (Dako Cytomation A/S) were used as the primary antibodies at dilutions of 1:3000, 1:100, and 1:100, respectively. Sections were lightly counterstained with methyl green. For negative controls, nonimmunized mouse IgG serum (Vector Laboratories, Burlingame, CA) was used as the primary antibody, and uniformly gave negative results. Stained sections were evaluated in a blinded manner without previous knowledge of the clinicopathologic parameters. Endothelial cells showed positive staining with a constant intensity; thus the staining intensity in the cytoplasm of the tumor cells was shown in comparison with that of endothelial cells and categorized as follows: weaker (level 1 VCP expression) or equal to or stronger than (level 2) that in endothelial cells. Cases showing combined level 1 and 2 staining in different areas of the tumor were classified as having level 2 expression. For p53 and Ki-67 immunohistochemistry, cells showing intranuclear staining were judged positive. The p53- or Ki-67-positive cells among 200 tumor cells were counted, and the percentage was taken as the p53 and Ki-67 labeling index. Cases were divided into two groups: for p53, level 1, p53 labeling index < 10%; level 2, p53 labeling index ≥ 10%; for Ki-67, level 1, Ki-67 labeling index < 20%; level 2, Ki-67 labeling index ≥ 20%.

Quantitative reverse transcription-PCR was performed on fresh-frozen samples from 12 patients with ESCC as described previously (20). Briefly, 10 μg of DNase I-treated total RNA were used for reverse transcription with Superscript III (Invitrogen, Carlsbad, CA). An aliquot representing 100 ng of input RNA was amplified by quantitative real-time PCR using the TaqMan PCR Reagent Kit and Assay-on-Demand Gene Expression Products (Applied Biosystems, Foster City, CA; Refs. 21, 22). RNA extracted from a noncancerous lesion in one patient was used as a standard. After reverse transcription, standard cDNA was serially diluted to obtain five standard solutions for use in PCR to generate the reference curve. The relative amount of cDNA in each sample was measured by interpolation using the standard curve (22), and then the relative ratio of VCP to β-actin expression was calculated for each ESCC sample.

Statistical analyses were performed with JMP software (SAS Institute Inc, Cary, NC). Correlation between the expression level of VCP determined by quantitative reverse transcription-PCR and immunohistochemistry was evaluated by a one-way ANOVA test. The χ² test and Fisher’s exact probability test were used to analyze the correlation between VCP expression as assessed immunohistochemically and the clinicopathologic features of ESCC. The Kaplan–Meier method was used to calculate survival rates (23), and differences in survival curves were evaluated with the log-rank test. Cox’s proportional hazards regression model with stepwise comparisons was used to analyze the independent prognostic factors (24). P values <0.05 were considered statistically significant.

Histologically, 44 tumors were well differentiated, 50 were moderately differentiated, and 61 were poorly differentiated squamous cell carcinomas. Tumor cells invaded the mucosa or submucosa (pT₁) in 38 patients, muscularis propria or subadventitia (pT₂) in 35, adventitia (pT₃) in 74, and adjacent organs (pT₄) in 8.

The 5-year disease-free and overall survival rates were 47.3% and 49.0%, respectively. Seventy-five patients showed tumor recurrence; recurrence was local in 18 patients, in the lymph nodes in 39, in the liver in 17, and in other organs in 14 patients. Sixty-six patients died from their tumors.

Three (1.9%) of 156 sections that did not show endothelial staining were regarded as having poor antigen preservation and were excluded from further analyses. The remaining 153 cases showing endothelial staining were evaluated for VCP expression. Cancer cells in 57 (37.2%) and 54 (35.3%) cases showed constant level 1 and level 2 VCP expression, respectively, in all areas of the specimens (Fig. 1). Tumor cells in 42 cases showed combined level 1 and 2 staining. In total, 96 cases (62.7%) were regarded as having level 2 VCP expression. Nontumorous esophageal mucosa constantly showed level 1 expression.

Quantitative reverse transcription-PCR analysis was performed on six ESSC patients with level 1 and six with level 2 expression. The relative ratios of VCP to β-actin expression in cases with level 1 and 2 expression were 0.4 ± 0.3 and 1.7 ± 1.4 (mean ± SD), respectively (P < 0.05; Fig. 2).

The correlation between VCP expression and clinical factors is listed in Table 1. Compared with ESCC with level 1 expression, cases with level 2 showed significantly higher frequencies of the following: presence of lymph node metastasis, 31 of 57 (54.4%) versus 72 of 96 (75.0%); and deep invasion (pT₃ and pT₄), 22 of 57 (38.5%) versus 59 of 96 (61.5%). We observed no correlation between VCP staining intensity and p53 labeling index or Ki-67 labeling index (Table 1). The correlation between VCP expression and patterns of tumor recurrence is shown in Table 2. Patients with level 2 ESCC showed higher frequencies of both local recurrence and distant metastasis than those with level 1 ESCC (P < 0.01 for both analyses).

Patients with level 1 ESCC had better 5-year survival rates than those with level 2 ESCC (disease-free, 65.9% versus 33.2%; overall, 68.7% versus 36.7%; P < 0.001 for both analyses; Table 3; Fig. 3). Univariate analysis revealed that tumor size, pattern of tumor growth, depth of tumor invasion, vascular invasion, lymphatic invasion, and lymph node metastasis were significant factors for both disease-free and overall survival (Table 3).

Multivariate analysis with factors shown to be significant in the univariate analysis revealed that VCP expression level in tumor cells, depth of tumor invasion, vascular invasion, and lymph node metastasis were independent prognostic factors for both disease-free and overall survival (Table 4).

The prognostic significance of VCP expression was further analyzed according to the pTNM classification (12). We found a significant difference in disease-free and overall survival between patients with level 1 and 2 expression at the early stage of disease (pT₁; P < 0.001 and P < 0.01, respectively; Table 5; Fig. 4). Lymph node metastasis was found in 4 (20.0%) of 20 level 1 patients and in 13 (72.2%) of 18 level 2 patients (P < 0.01). At the advanced stage (pT₂–pT₄), we observed a significant difference between patients with level 1 and 2 expression for disease-free and overall survival (P < 0.05 for both analyses; Fig. 4).

The conventional TNM staging system for ESCC provides useful information for the prediction of tumor recurrence and patient survival (5, 6, 11). Recent advances in therapeutic modalities for ESCC, however, have yielded new problems to be solved. For example, recently introduced reduction surgeries, such as endoscopic mucosal resection and transhiatal esophagectomy, enable tumor excision without a severe influence on the patient’s general condition (25, 26, 27, 28); however, lymph node metastasis, observed at relatively high frequencies even in early ESCC (29), becomes a risk factor when these techniques are used. In contrast, the introduction of adjuvant chemo- and radiotherapy has improved the prognosis of patients with ESCC, particularly those with high potential for lymph node metastasis (30, 31). Under these circumstances, preoperative assessment of lymph node metastasis and the metastatic potential of ESCC is essential for therapeutic decision-making.

In the present study, we examined the correlation of VCP expression with the previously proposed prognosticators, p53 expression and proliferative activity shown by Ki-67 immunohistochemistry, and found no relationship between them, demonstrating that VCP expression is a prognosticator independent of previously known factors.

The patient characteristics, such as gender, age, primary site, and 5-year survival rate, in the present series were similar to those in previous reports from Japan and other Asian countries where ESCC is common (1, 10, 15, 29). ESCC in Western countries, where the disease is relatively uncommon, shares the above-mentioned characteristics except that tumors are rather frequent in the lower esophagus (5, 6). In addition, the uni- and multivariate analyses in the present study showed the prognostic significance of clinicopathologic factors such as size and depth of the tumor and lymph node metastasis, as reported previously from Western countries (5, 9) and Japan (10). These findings indicate that the results obtained from the present series of cases are applicable to ESCC in other counties.

VCP is involved in the ubiquitin/proteasome-dependent protein degradation pathway, which plays an essential role in controlling the levels of various cellular proteins and therefore regulates basic cellular processes such as cell cycle progression, signal transduction, and cell transformation (17). Notably, VCP is involved in regulating the activation of nuclear factor-κB (16, 17), a transcription factor whose activity is correlated with antiapoptosis, cell proliferation, and invasion (32). It is therefore postulated that VCP plays a crucial role in tumor invasion and metastasis.

We examined VCP expression in 12 cases at the mRNA and protein levels by quantitative reverse transcription-PCR and immunohistochemical analyses, respectively, and found a clear correlation between the two measurements. The present results, together with our previous study on hepatocellular carcinoma and prostate cancer (20, 33) and the study by Muller et al. (34) on murine tissues, indicates the reliability of immunohistochemistry in the measurement of VCP expression.

Among the clinicopathologic factors examined, we observed a significant correlation between VCP expression and depth of invasion and lymph node metastasis, indicating the close association between VCP expression and growth and invasiveness of ESCC. These findings are consistent with our reports of the correlation of VCP overexpression with increased metastatic potential of tumor cells in an experimental metastasis model (16) and increased recurrence and poor prognosis of hepatocellular carcinomas in a clinical analysis (20).

The uni- and multivariate analyses revealed that the VCP expression level is an independent prognosticator for ESCC recurrence and patient survival. In addition, the assessment of VCP expression is useful for predicting the recurrence of ESCC in patients in both the early (pT₁) and advanced (pT₂–pT₄) stages: the 5-year disease-free survival rates in patients with VCP level 1 and 2 were 87.5% and 29.9% at stage pT₁ and 56.7% and 33.4% at stages pT₂ through pT₄, respectively. The combination of pTNM classification and VCP expression level in tumor cells is useful for predicting the prognosis of patients with ESCC.

Grouping the cases with ESCC based on the present system could be a good guide for choosing the various modalities of adjuvant therapy. In patients with VCP level 1 ESCC in the early stages, a favorable outcome could be expected. Lymph node metastasis was observed in only 20% of these patients; therefore, reduction surgeries might be considered, particularly for those in poor general condition or with a small tumor. In contrast, patients with level 2 ESCC and/or advanced stage disease might have a dismal prognosis, and the introduction of intensive adjuvant therapies for these patients might be justified.

In conclusion, VCP expression as determined by immunohistochemistry could be a new prognostic marker for ESCC. The present study indicated that the stratification of ESCC patients based on the disease stage and VCP expression level would be valuable for predicting tumor recurrence and prognosis. This system might provide a novel way to explore effective treatment modalities for ESCC.