Cell Culture in Esophageal Squamous Cell Carcinoma and the Association with Molecular Markers¹

The establishing of cell lines from tumor samples has been thought to be an indicator of biological malignancy. However, the use of cell culture techniques has been limited by the low success rate of in vitro culture; hence, the clinical application of the cell culture growth potential has also been limited. There have been few reports published concerning the growth and behavior of human esophageal carcinoma cells in culture (1, 2, 3). We revealed previously that the establishment of cell lines was a significant prognostic factor in esophageal cancers (4); however, the sample size was too small to evaluate the association between the establishment of cell lines and the genetic abnormalities at that time.

After previous observations, we have continued to establish esophageal cancer cell lines and succeeded to establish >45 cell lines of ESCC³ (5). These cell lines had several gene abnormalities; p53 mutation (6), p16 methylation (7), cyclin D1 overexpression (8), p21 expression (9), FHIT methylation (10), MDM2 overexpression (11), and vascular endothelial growth factor expression (12). We also checked p27 status in primary ESCC (13). From these observations, the following questions were raised: (a) was the establishment of esophageal cancer cell lines a real independent biological prognostic factor of the ESCC; and (b) what kind of genetic abnormalities contributed to the establishment of carcinoma cell lines. To solve these problems, we retrospectively evaluated the association with the establishment of esophageal carcinoma cell lines and the genetic abnormality. To clarify which factors are the true contributing factors in the establishment of cell lines we examined the expression of the following six selected molecular biological factors: p53, p21/Waf1, p27/Kip1, MDM2, cyclin D1, and FHIT. This time, for ease in clinical application and in obtaining the primary tumors, we carried out an immunohistochemical analysis.

From 1986 to 1998, 298 esophagectomies for esophageal cancer were performed in our department, and 243 esophageal resected primary tumors were used to attempt to establish a cancer cell line. Two hundred and three of the 243 patients underwent a curative esophagectomy (R0) and were used in this study. There were no operative deaths. All of the patients had pathological SCCs, and the clinical status of each patient was classified according to the pTNM classification system (14). There were no patients with distant organ metastasis, and all of the stage 4 patients had distant lymph node metastasis. The mean age was 64.1 ± 10.1 years (Table 1).

The standard surgical method that was used has been described previously (15). In brief, esophagectomies with lymph node dissection were performed by a right thoracotomy, and subsequent reconstructions were carried out by an esophagogastrostomy using a gastric tube through the retrosternal route. Sixteen patients (7.9%) received preoperative irradiation (20–40 Gy), and 76 patients (37.4%) received preoperative chemotherapy (mainly CDDP 50 mg × 2; Ref. 16). Seventy-nine patients (38.9%) received postoperative irradiation (50.4–60 Gy; Ref. 17), and 36 patients (17.7%) received postoperative chemotherapy (mainly CDDP 50 mg × 2; Ref. 16). The minimal follow-up period of the surviving patients in this study was 2 years (mean 6.3; SD 3.8). Written informed consent was obtained from the patients for surgery and to use resected samples for research. The approval numbers of Institutional Review Board of Kyoto University were #232 and G48.

Cells from resected esophageal cancers were cultured using the techniques proposed by Shimada et al. (4, 18). Culture samples were obtained from resected specimens in the operating room. Details of the culture methods and cell characterization have been described previously (4, 18). In brief, the samples were taken from the mucosal surface and were washed four times with MEM containing 10 μg/ml of amphotericin B. Necrotic or hemorrhagic regions, associated muscle, and fat were removed. The average size of the tumor samples taken from the resected esophagi was 1.0 cm³. The cancer tissue was minced with sharp scissors into pieces of ∼0.5 mm in diameter, and a crude cell suspension was made by pressing the minced tissue through a stainless mesh. This cell suspension was dispersed into 60-mm cell culture dishes along with 6 ml of culture medium for primary culture, and covered with a 90-mm Petri dish to avoid inversion. Ham’s F12/RPMI 1640 containing 20% FCS, penicillin (100 units/ml), and gentamicin (2.5 μg/ml) were used for the primary culture. The cultures were maintained at 37°C in a humidified atmosphere containing 5% carbon dioxide until confluent. The cultures were examined daily, and half of the medium volume was usually renewed twice a week. Fibroblast overgrowth was controlled by differential trypsinization and scraping. Initial cell passage was performed when rapid tumor cell growth was observed. After the cells had adapted to culture conditions they were maintained in Ham’s F12/RPMI 1640 with 2% FCS.

If continuous cell growth was obtained over 2 years or cells could undergo >50 passages, these cells were defined as cell lines. The resected esophageal samples were cultured in Petri dishes, and culture patterning was checked under phase contrast microscopy as described previously (4, 18). Cases in which <3 culture dishes were obtained were excluded from this study, because too small a sample might reduce the rate of establishment.

Immunohistochemical analysis was done retrospectively. Resected esophageal specimens, which included both tumor and normal mucosa, were fixed in a 10% formaldehyde solution and embedded in paraffin. Four-μm sections were cut and mounted on APS coated glass slides. Immunohistochemical staining was performed using the avidin-biotin method as described previously (12).

The unmasking of antigens was carried out by incubation in citrate buffer (pH 6.0) at 95°C for 10 min (microwave or autoclave). Sections were incubated overnight at 4°C with each antibody in PBS containing 1% BSA and were counterstained with Mayer’s hematoxylin. Antirabbit or antimouse IgG was used for checking nonspecific staining.

The intensity of immunohistological staining was evaluated in five areas of the slide section for correlation and confirmation of the tissue analysis. All of the samples were evaluated by two of the authors (Y. S. and M. M.). When the results of the two investigators were not the same, a final consensus was obtained through careful discussion.

For p53 and cyclin D1, >10% positive staining in nuclei was defined as positive staining (19, 20).

For Mdm2, >30% positive staining in nuclei was defined as positive staining (21).

For p21/Waf1 and p27, >50% positive staining in nuclei was defined as positive staining (13, 22).

For FHIT, strong staining the same as normal epithelia or lymph node was classified as positive staining (23).

The representative staining of each antibody was demonstrated in Fig. 1.

Ham’s F12 and MEM were purchased from Nissui Corporation, Ltd. (Tokyo, Japan), RPMI 1640 was purchased from Life Technologies, Inc. (Grand Island, NY), and FCS was purchased from ICN Biomedical, Inc (Aurora, OH).

The antibodies used in this study were as follows: p53 (DO-7; DAKO, Tokyo, Japan), p21/Waf1 (Waf1 Ab-1; Oncogene Science, Cambridge, MA), cyclin D1 (anti-CyclinD1; Novocastra, Newcastle, United Kingdom), p27 (anti-p27; Transduction Laboratories, Lexington, KY), Mdm2 (anti-Mdm2; Novocastra), and FHIT (F-130; IBL, Gumma, Japan).

Positive stained cases were designated “1” and negative stained cases were designated “0.” The establishment of cell lines, pT, pN, pM, histological grade, age, and gender were divided into the following two categories: (not establishment of cell line: 0; establishment of cell line: 1; pT1, pT2: 0; pT3, pT4: 1; pN0: 0; pN1: 1; pM0: 0; pM1: 1; G1, G2: 0; G3, G4: 1; <64 years old: 0; ≥64 years: 1; female: 0; male: 1), With regard to pre- and postoperative treatment, terms were simply divided into the following two categories: (not done: 0; done: 1).

The age of the patients was compared by a t test. Survival curves of the patients were calculated by the Kaplan-Meier method and analyzed by the log-rank test. Multivariate analysis was performed using Cox’s proportional hazard model and logistic regression analysis (24). The correlation between cell culture and molecular parameters was statistically evaluated using Fisher’s exact test or χ² test. All tests of statistical significance were evaluated by two-sided analysis. The software used was the JMP version 4 for Macintosh (SAS Institute Inc., Cary, NC).

Among the 203 R0 resected esophageal cancer specimens, 35 cell lines were established with a success rate of 17.2%. Of the 35 cell lines, 7 were derived from patients with well-differentiated SCCs, 19 from those with moderately differentiated SCC, and 9 from those with poorly differentiated SCC. The clinical and pathological backgrounds of the patients are shown in Table 2: group A (n = 35), from whom cancer cells could be cultured as continuous cell lines, and group B (n = 168), from whom cell lines could not be established. The depth of tumor (pT) and lymph node metastasis were associated with the establishment of cell lines (P < 0.05). There were no significant differences between the two groups with regard to sex, age, histological type, histological stage, preoperative treatment, or postoperative treatment.

First, we analyzed the prognostic impact of cell culture. The cumulative survival rate of patients in the cell line group was significantly lower than that of those in the noncell line group (P = 0.0006; Fig. 2). The 5-year survival rate was 45.9% in the noncell line group, whereas it was 19.9% in the cell line group. Furthermore, Cox’s proportional hazard model revealed that cell culture capability was an independent prognostic factor (RR, 1.98; P = 0.007) and pN was the strongest (RR, 3.04; P < 0.0001; Table 3). Pre- and postoperative treatment had no effect on patient survival. Even if the patients divided into “any chemo- or radiotherapy” versus “no chemo- or radiotherapy,” chemo or radiotherapy had no effect on patient survival (RR, 0.87; 95% CI, 0.54–1.43; P = 0.577).

Next, we evaluated the predictive factor of the establishment of the cell line. Univariate logistic regression analysis revealed that cell culture capability had associations with the following pathological and molecular biological factors (Table 4): pN, cyclin D1, p53, MDM2, p27, and FHIT (P < 0.05). However, multivariate logistic regression analysis revealed that p53 protein accumulation and MDM2 protein expression predict establishment of cell line in ESCC (OR, 7.72 and 8.62, respectively; Table 5).

Finally, we evaluated the association among cyclin D1, p53, MDM2, p27, and FHIT in ESCC. There was a significant association between cyclin D1 expression and MDM2 expression (P = 0.039); however, expression of cyclin D1 had no significant association with p53 accumulation (P = 0.41).

There are >100 esophageal cancer cell lines (5). However, many cell lines have already been contaminated and are of no use for medical research. We have reported previously that the patients in whom cancer cells could be cultured as continuous cell lines had poor prognosis in ESCC patients using 50 esophageal cancer patients and 21 esophageal cancer cell lines (4). At that time, we used DNA fingerprint analysis to check the cell line individuality. After that, we have established >45 esophageal cancer cell lines and recent analysis of comparative genomic hybridization confirmed that there was no cross-contamination in our cell lines (25, 26). Therefore, we can perform additional analysis of the prognostic impact of the establishment of cell lines.

In this study, univariate and multivariate proportional hazards model revealed that the establishment of carcinoma cell lines was an independent prognostic factor. Whereas this growth property correlated significantly with the incidence of lymph node metastasis, there was no correlation with histopathological differentiation. All of the surgical samples were taken from resected esophagus, thus it would seem that the ability of cells to grow continuously in vitro depends on the characteristics of the primary lesion and not on metastatic lesions such as in lymph nodes. Accordingly, cancer cells, which grew continuously in vitro, were inclined to metastasize to lymph nodes.

Preoperative treatment had been thought to reduce the establishment of cancer cell lines; however, in this series, preoperative low doses of cisplatin and preoperative irradiation (40 Gy) did not affect the establishment of cancer cell lines. Although preoperative treatments were performed in advanced cases, the treatments did not have enough effect to reduce the progression of cancer cells. Multivariate analysis, which included the preoperative treatment, also revealed that preoperative treatment had no effect on the prognosis of the patients.

Our results suggested that p53 and its regulation gene MDM2 have an important role in the establishment of cell lines. Interestingly, in the fact of the establishment of cell lines, our results revealed that the status of p53 and MDM2 was more important than the status of lymph node metastasis. Our previous results did not show the association between p53 mutation and establishment of cell lines (6); however, the number of patients was too few to obtain a definite conclusion at that time. The MDM2 gene, which was originally thought to be related to sarcomatous tumors, was associated with a worse prognosis in esophageal cancer (11). The product of the MDM2 gene is known to bind to p53 protein and inhibit its ability to activate transcription (27, 28). It is suggested that some aspects of regulation of cellular proliferation by p53 can be abrogated by MDM2 (27). When the MDM2 gene is amplified and overexpressed, extended suppression of the function of the p53 protein may result in uncontrolled cell growth. There has been a recent revelation that the MDM2 gene also regulates the retinoblastoma gene (29, 30). These results suggested that p53 and the regulated genes abnormalities provide an optimal condition that can survive and grow in the cell culture system to the esophageal cancer cells.

FHIT protein has been characterized as an AP3A hydrolase, which cleaves the AP4A substrate, a molecule that might be involved in the control of DNA replication and cell cycle (31). We reported previously that FHIT protein expression associated with tumor progression (23), and the present study also indicated that FHIT protein has some role in the establishment of esophageal cancer cell lines (OR, 8.62). However, multivariate analysis did not show independent contribution to the establishment of esophageal cancer cell lines.

Unlike P53 and FHIT protein, p21 and p27 protein expression had a positive risk factor in the establishment of cell lines (OR, 2.61 and 1.5, respectively). This means that p53 and FHIT protein act as tumor suppressors of in vitro culture; however p21 and p27 had less effect than p53 and FHIT.

Overexpression of cyclin D1 has been suggested to contribute to oncogenesis by disturbing the cell cycle (32, 33). D cyclin binds to the cyclin-dependent kinases (CDK2 and CDK4) and to proliferating cell nuclear antigens, and formation of these complexes has been implicated in the control of cell proliferation (34). Therefore, it is likely that an amplification and overexpression of the cyclin D1 gene could lead to uncontrollable cell growth and the proliferation of tumor cells. Contrary to these observations, our present study could not show the independent contribution to the establishment of esophageal cancer cell lines. However, immunohistochemical staining detected only the expression of cyclin D1; thus, the present study might not represent the overexpression of cyclin D1. Furthermore, there was the possibility that cyclin D1 became nonsignificant because of its association with MDM2.

We used immunohistochemical analysis to evaluate p53 status; however, there are some criticisms regarding this detection method. Although p53 staining did not always correlate with the results of direct sequencing (35) and many reports have already indicated that p53 expression has heterogeneity, the relationship between p53 protein accumulation and p53 mutation, there was concordance between p53 mutation and protein expression [69% (36) to 76.2% (37)]. Therefore, our data did not indicate sequence abnormalities of the p53 gene, but only indicated that p53 protein accumulation was associated with the establishment of cell lines.

There is also another criticism as to whether the genetic status of the establishment of cell lines is reflected by primary tumors. Our previous study of p53 mutation revealed that in the 7 cell lines established from mutation-free tumors, newly acquired mutations were detected in 5, which suggested that mutations might occur during the process of establishing cell lines (6). We also revealed previously that the inactivation of the p16 gene in primary tumors was not as frequent as that of the p16 gene in 30 cell lines (7). On the other hand, MDM2 amplification in the cell lines was the same as the original tumor (11). These previous observations indicated that genetic analysis (especially of tumor suppressor genes) of cell lines did not always reflect the genetic status of primary tumors. These genetic changes may be either permissive or required for in vitro growth. Therefore, the primary tumor does not always predict cell line capability.

In conclusion, we reconfirmed the prognostic significance of cell culture in ESCC, and revealed that p53 protein accumulation and MDM 2 protein expression predict establishment of esophageal cancer cell line. Our results suggested that p53 and MDM2 had a crucial role in the establishment of cell culture.

We thank Sakiko Shimada for maintaining the cell culture. We also thank Toby Cavanaugh for support in proofreading the manuscript and Dr. Shunzou Maetani, Vice President of the Research Center of Tenri Hospital, for excellent advice in statistical analysis.