Spontaneous Apoptosis and the Expression of p53 and Bcl-2 Family Proteins in Locally Advanced Head and Neck Cancer

Background  Apoptotic cell death plays a key role in the pathogenesis, aggressiveness, and therapy responsiveness of cancer. The suicidal machinery of apoptosis is genetically controlled. Proteins of the Bcl-2 family as well as p53 are important regulators of apoptosis.

Objective  To assess the rate of spontaneous apoptosis and the expression of p53 and Bcl-2 family proteins in locally advanced squamous cell carcinomas of the head and neck.

Design  Twenty-six patients with locally advanced squamous cell carcinoma of the head and neck were included in the study. The expression of p53, Bcl-2, Mcl-1, Bax, and Bak was assessed by immunohistochemical analysis. The terminal deoxytransferase-mediated deoxyuridine nick end-labeling assay was used to quantify apoptosis by flow cytometry.

Results  Tumor cells containing immunostaining for the antiapoptotic proteins Bcl-2 and Mcl-1 were present in 4 (15%) and 24 (92%) of the cases evaluated, respectively, whereas immunopositivity for the proapoptotic proteins Bax and Bak was found in 9 (35%) and 24 (92%) of the samples. Immunoreactivity to p53 was detected in 20 (77%) of the samples. There was a positive correlation between the expression of Bcl-2 and Bax and between Mcl-1 and Bak. A low fraction of apoptotic cells (<2.5%) in the pretreatment tumor samples was significantly correlated with increased 2-year survival in these patients.

Conclusions  Our results establish the frequent expression of the Bcl-2 family proteins Bcl-2, Mcl-1, Bax, and Bak in locally advanced head and neck cancer. In addition, this study suggests that the apoptotic fraction in pretreatment tumor samples might be of prognostic importance for the outcome in these patients.

SQUAMOUS CELL carcinoma of the head and neck (SCCHN) accounts for 5% of all adult cancers worldwide. Patients with stage I and II disease¹ can be effectively treated with surgery or radiotherapy and achieve a 2-year disease-free survival of 60% to 90%. However, two thirds of the tumors are already advanced (stages III and IV) at the time of the initial examination, and only 40% to 50% of these patients survive 2 years if treated with surgery alone or in combination with radiotherapy. Two thirds of the recurrences occur locally, while one third of the patients develop distant metastases. In case of a locoregional recurrence, the chance of cure or of a durable palliation is less than 10% with additional local therapeutic measures. Survival rates for patients with advanced SCCHN have remained almost unchanged in the last 30 years. This situation makes the availability of prognostic factors for outcome highly desirable to guide the use of (neo)adjuvant treatment modalities in the context of clinical studies and facilitate deciding about the extent of the initial operative procedure.

Recently, deranged regulation of apoptotic cell death has evolved as an important factor not only for the pathogenesis of cancer but also for its biological behavior and the responsiveness to chemotherapy and radiotherapy.^2,3 Various hierarchical levels of genes are involved in the control of apoptotic cell death. Sensors such as the tumor suppressor p53 can trigger apoptosis as a reaction to unfavorable conditions. More downstream, the interplay of different Bcl-2 family members determines whether apoptosis is allowed to proceed. At the executive end of this hierarchy, a whole cascade of caspases is involved to effect the typical morphologic changes of apoptosis. The interplay of all these gene products and other factors determines the threshold for the induction of apoptosis in an individual cell.² Cells with a high apoptosis threshold tend to be more resistant to environmental stresses such as anaerobic conditions, the lack of growth factors, or the action of chemotherapeutic drugs.⁴ Thus, the apoptosis threshold is an important determinant for the biological characteristics of normal cells as well as cancer cells.

Among the regulators of cell death, Bcl-2 (B-cell lymphoma/leukemia-2) stands out for its evolutionary conservation and its ability to regulate a highly downstream step of the cell death pathway.^5,6 High levels of Bcl-2 protein can protect cancer cells from apoptotic cell death induced by a wide variety of stimuli and insults, including radiation and nearly all chemotherapeutic drugs. The discovery of genes encoding proteins with substantial amino acid sequence homology has led to the expanding family of Bcl-2–related genes.⁵ The composition or ratio of the different Bcl-2 family members may play an important role in the pathogenesis of cancer and determine the chemosensitivity in an individual tumor. The molecular details of these interactions are not fully understood. Some of the Bcl-2 family members, such as Mcl-1 (myeloid cell leukemia-1), seem to block cell death, while others, such as Bax or Bak, abrogate Bcl-2 function and promote cell death.⁵ In head and neck cancer, only the prognostic importance of Bcl-2 expression has been examined to date, and the results of these studies are controversial.^7-11

Mutations of the p53 tumor suppressor gene belong to the most frequent gene alterations in solid tumors.¹² Loss of p53 function may partly explain their notorious resistance to chemotherapy and radiotherapy, since DNA damage cannot be sensed and translated into apoptosis.^2,4 In addition, p53 mutations also can lead to gain of function, conferring increased tumorigenicity, proliferative capacity, metastatic potential, and tissue invasiveness.¹² All of these properties should predict a poor clinical outcome with tumors containing mutant p53, but this association is controversial for head and neck cancer.^13,14

The present study assessed the differential expression pattern and the prognostic importance of the apoptosis-regulating proteins p53 and the Bcl-2 family members Bcl-2, Mcl-1, Bax, and Bak in patients with locally advanced SCCHN. As a more direct measure of apoptosis, the terminal deoxytransferase-mediated deoxyuridine nick end-labeling (TUNEL) reaction was performed on ground tumor tissues and quantified by flow cytometer analysis.

Patients who were treated for T3 and T4 SCCHN without distant metastasis between January 1, 1993, and December 31, 1995, at the Institut Gustave-Roussy, Villejuif, France, and the Otorhinolaryngology Department of the University Hospital Bern, Bern, Switzerland, were eligible for this study if sufficient material could be collected for the planned analyses. The standardized staging procedure contained radiological evaluation and diagnostic panendoscopy with biopsy. All patients were followed up every 1 to 3 months until death.

Paraffin-embedded tumor specimens that had been fixed in neutral-buffered formalin were sectioned (4 mm) and immunostained with antipeptide polyclonal antisera specific for Mcl-1 (1:800), Bax (1:2000), and Bak (1:1000) as previously described in detail.¹⁵ The specificity of all these antibody reagents had been demonstrated on the basis of comparisons with preimmune serum, peptide competition experiments, and immunoblot analyses of tissue extracts and cell lines.¹⁵ Monoclonal antibodies (both from Dako, Glostrop, Denmark) were used for p53 (M 7001, DAKO-p53, DO-7; 1:50) and for Bcl-2 (MO 887, DAKO-Bcl-2, 124; 1:50). Immunodetection was achieved by an avidin-biotin horseradish-peroxidase–based colorimetric method with 3,3‘-diaminobenzidine used as the chromogen, followed by light counterstaining with hematoxylin. The slides were assessed for each antibody by 2 reviewers unaware of patient characteristics and outcome (M.A.H., M.M.B.). Specimens were considered immunopositive when more than 5% of the tumor cells had clear evidence of immunostaining. Negative controls were processed by exclusion of the primary antibody. The positive internal controls are described in the "Results" section.

A part of the fresh tumor sample was processed to a single-cell suspension. Immediately after removal, the tumor tissue was minced either manually or with a special tissue grinder (Medimachine; Dako). Cells were fixed in 1% formaldehyde at pH 7.4 on ice for 15 minutes, centrifuged, and postfixed in 70% ethanol at −20°C after 1 washing step in Hanks balanced saline solution. To stain apoptotic cells, DNA strand breaks were labeled by means of TUNEL.¹⁶ After washing and fixation, cells were processed with an in situ apoptosis detection kit (ApopTag) according to the instructions of the manufacturer (Appligene/Oncor, Illkirch, France). Flow cytometry was performed on a flow cytometer (FACScan; Becton Dickinson & Co, Mountain View, Calif), and cell populations were quantified with CELL QUEST software (Becton Dickinson & Co). Apoptosis was morphologically confirmed by fluorescent microscopy.

Pearson correlation was used to compare immunostaining results (p53, Bcl-2, Mcl-1, Bax, Bak, and apoptosis) with other categorical features (tumor stage, tumor site, age, and therapy). An apoptotic score cutoff value of 2.5% was chosen by dividing the population into equally sized groups. Actuarial overall survival rates were calculated from the date of the definitive tumor operation to death induced by local relapse or metastasis. All 26 patients had a 2-year minimum follow-up. To assess the influence of the above-mentioned factors on the hazard function for survival, the proportional hazard regression model in combination with backward elimination was used. The overall survival curves were estimated according to the Kaplan-Meier method.¹⁷ Survival differences were evaluated with the log-rank test. All tests were 2 sided.

A total of 26 samples of advanced SCCHN were processed for apoptosis and expression of p53, Bcl-2, Mcl-1, Bax, and Bak. Table 1 lists patient and tumor characteristics. Ten (38%) of 26 patients developed distant metastasis and 10 (38%) of 26 had a local progression or relapse. After a 24-month follow-up, 13 of 26 patients died of their disease. Three patients died of metastases, 2 of local progression, and 8 of combined local and metastatic disease. Table 2 lists the frequency of positivity for the various markers and the percentage of spontaneous apoptosis in the tumor samples of these patients. The immunostaining for p53 was nuclear (Figure 1, A), while staining for Bcl-2 family proteins was typically cytoplasmic, since these proteins are usually associated with cytosolic organelles such as mitochondria and endoplasmic reticulum.

The monoclonal antibody DO-7 specifically detects the wild-type and mutant forms of p53. Positive immunostaining for p53 suggests mutant p53, since normal p53 is virtually undetectable by antibody-mediated staining because of its rapid degradation.¹² Immunoreactivity for p53 was found positive in 20 patients (77%) (Figure 1, A) and negative in 6 patients (23%). Nine (45%) of 20 patients with p53-positive tumors survived, compared with 4 (67%) of 6 patients with negative p53 staining. There was no significant correlation between p53 accumulation and the expression of Bcl-2 family proteins (Pearson coefficient of correlation, r=−0.2).

Immunoreactivity for Bcl-2 was found in 4 patients (15%) (Figure 1, B). The basal layers of the normal epithelium or mantle zone lymphocytes, where normal to strong Bcl-2 immunostaining was typically seen, served as a positive control. All 4 patients with positive results survived for more than 2 years. There was a positive correlation between Bcl-2 accumulation and the expression of Bax (Pearson coefficient of correlation, r=0.55).

Immunoreactivity for Mcl-1 was positive (Figure 1, C) in 24 patients (92%). Chondrocytes and germinal center lymphocytes were used as positive internal controls for Mcl-1.¹⁸ There was a positive correlation between Mcl-1 accumulation and the expression of Bak (Pearson coefficient of correlation, r=0.8).

Immunoreactivity for Bax was positive (Figure 1, D) in 9 patients (35%). Interfollicular lymphocytes served as an internal positive control for Bax expression.¹⁹

Immunoreactivity for Bak was positive (Figure 1, E) in 24 patients (92%). Both patients with negative results survived the 2 years. Smooth muscle cells of arteries served as an internal positive control for Bak expression.²⁰

The percentage of apoptotic cells varied from 0.2% to 13.4% (Table 2). The actuarial survival of patients with a low (≤2.5%) fraction of apoptotic cells in their tumors was significantly longer (log-rank P=.001) than that of patients with an apoptotic fraction greater than 2.5% (Figure 2). With the use of the proportional hazard regression model with backward elimination as described in the "Statistical Analysis" section, apoptosis was confirmed as an independent prognostic factor for survival with a hazard ratio of 9.6 (95% confidence interval, 2.1-44.1; P=.004). No significant correlation between the expression of apoptosis-regulating genes and the occurrence of spontaneous apoptosis was found.

The deranged regulation of apoptotic cell death has evolved as an important factor for the biological behavior of cancer.^2,3 Different Bcl-2 family members and p53 are involved in the control of the apoptosis machinery, and this is, to our knowledge, the first study to examine the expression pattern of Bax, Bak, and Mcl-1 in addition to Bcl-2 and p53 in locally advanced head and neck cancer. As a more direct measure of apoptosis, the TUNEL assay was performed on single-cell suspensions of tumor tissue and quantified by flow cytometer analysis.

Mcl-1, Bak, Bax, and Bcl-2 were frequently expressed, with 92%, 35%, 92%, and 15% of the tumors staining positive, respectively. At the protein level, the various Bcl-2 family members can dimerize with one another, with one monomer antagonizing or enhancing the function of the other. In this way, the ratio of inhibitors to activators may determine the propensity of a cell to undergo apoptosis. With the exception of Bcl-2, relatively little is known about the expression of other family members in human cancer. In a series of 64 patients with prostate cancer, Bax was expressed in all samples, while Mcl-1 expression was found in 81% and Bcl-2 in 25% of the cases.²¹ In stomach cancer, the antiapoptotic proteins Bcl-2 and Mcl-1 were present in 54% and 75% of the cases evaluated, while the proapoptotic proteins Bax and Bac were found in 92% and 88% of the specimens, respectively.²¹ In all 3 series including ours, Bcl-2 has shown the most restricted expression of all family members. Since one function of Bcl-2 is to provide protection from apoptosis, these results could be taken as evidence of a low apoptosis threshold of these tumors. However, this is contradicted by their notorious chemoresistance. Thus, it is possible that the role of Bcl-2 is redundant and that other antiapoptotic members of the Bcl-2 family can take over its function, as suggested by the reciprocal pattern of Bcl-2 and Mcl-1 expression in normal tissues.¹⁸ Interestingly, in our series we found a positive correlation between the expression of apoptosis-promoting proteins (Bax, Bak) and apoptosis-suppressing proteins (Bcl-2, Mcl-1). A similar pattern has been found in Reed-Sternberg cells of Hodgkin disease, suggesting that Bcl-2 or other family members such as Mcl-1 may neutralize the cell death–promoting activity of Bax, allowing malignant cell survival.²² Since transformation per se seems to lower the apoptosis threshold, apoptosis-suppressing activity should be a requirement for the occurrence of cancer.²

Mutations of p53 are frequent in SCCHN and are considered an early event in carcinogenesis.^14,23 We found p53 expression in 77% of our tumor samples, and this result concurs with other data from the literature.¹⁰ Positive immunostaining for p53 in clinical samples appears to be associated with p53 mutations, since the half-life of the defective protein is prolonged.¹² Forty-five percent of our patients with p53-positive tumors survived, compared with 67% with p53 negative cancers. However, this difference was not statistically different. In the series by Spafford et al,⁸ p53 expression was correlated with a decreased survival. A higher incidence of local relapse and a faster progression from in situ to invasive SCCHN lesions was found in p53-positive patients by Munck-Wikland et al.²⁴ All these results indicate that the expression of mutant p53 is of biological significance in SCCHN.

By dividing the patient population into 2 equally sized groups based on the apoptotic index, it was possible to define a cutoff point at 2.5% apoptotic cells in our series, which separated a good-prognosis group of patients with a low fraction of apoptotic cells from a poor-prognosis group with a significantly reduced life expectancy after local treatment for locally advanced SCCHN. Apoptosis was confirmed as an important prognostic factor in the proportional regression model. There was no significant correlation between the expression of apoptosis-regulating genes and the occurrence of spontaneous apoptosis in fresh tumor samples. This confirms the complexity of apoptosis control by an expanding number of putative regulators. The number of major factors involved seems to make it unlikely that a straightforward relationship exists between a single regulator protein and clinical end points. However, our results are internally consistent, since the expression of none of the apoptosis-regulating proteins was associated with survival or apoptosis, whereas apoptosis and survival were related.

Instead of verifying the occurrence of apoptosis in tissue sections, we used flow cytometry to detect apoptosis in fixed single-cell solutions. This method allows optimal cell fixation to prevent DNA leaking out of the cell and simple quantification of the apoptotic events. The percentage of apoptotic events measured in the tumor samples by TUNEL may not only represent the rate but also the duration of apoptosis in the tumor. Until now no cumulative estimate of the rate of cell entrance to apoptosis exists as does for mitosis, which can be arrested and assessed by microtubule poisons in stathmokinetic experiments.²⁵

At this point, pathological staging remains the most reliable determinant of prognosis in SCCHN and the main factor in the choice of curative treatment. However, it is of great interest to verify new biological markers to define the risk of relapse or to decide on the use of (neo)adjuvant treatment. In this context, our observation of an inverse correlation between the apoptotic fraction and outcome is of potential interest and requires confirmation.

Accepted for publication September 10, 1998.

This research was supported by grants from the Swiss Cancer League, the Bernische Stiftung für klinische Krebsforschung, a donation from the Berner Männerchor, and the Grant in Aid Fund of the University of Bern, Bern, Switzerland.

Reprints: Michel A. Hotz, MD, Clinic of Otorhinolaryngology, Head and Neck Surgery, Inselspital, 3010 Bern, Switzerland (e-mail: michael.hotz@insel.ch)