Expression of p21WAF1/CIP1 in relation to survivors and nonsurvivors in 80 patients with tongue squamous cell carcinoma. The horizontal line indicates the 23% cutoff level (P = .03).
Expression of p21WAF1/CIP1 in relation to survivors and nonsurvivors in 33 cases of p53-negative tongue squamous cell carcinoma. The horizontal line indicates the 23% cutoff level (P = .002).
Disease-specific survival curves (Kaplan-Meier) for p21WAF1/CIP1 expression in 80 patients with tongue squamous cell carcinomas (P = .04).
Disease-specific survival curves (Kaplan-Meier) for the p53 − p21WAF1/CIP1 combination score in 80 patients with tongue squamous cell carcinomas (P<.001).
Disease-specific survival curves (Kaplan-Meier) for the Bax + p21WAF1/CIP1 combination score in 80 patients with tongue squamous cell carcinomas (P<.001).
Xie X, Clausen OPF, Boysen M. Prognostic Significance of p21WAF1/CIP1 Expression in Tongue Squamous Cell Carcinomas. Arch Otolaryngol Head Neck Surg. 2002;128(8):897-902. doi:10.1001/archotol.128.8.897
Copyright 2002 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2002
To investigate the prognostic significance of p21WAF1/CIP1 expression and its relationship with p53 accumulation and other apoptotic markers such as Bax, Bcl-2, and apoptotic index in relation to disease-specific survival in oral tongue squamous cell carcinoma (SCC).
A 10-year retrospective clinical study. Information about clinical findings, treatment, and follow-up has been recorded prospectively.
Patients and Methods
Diagnostic, formalin-fixed, paraffin-embedded sections taken from 80 randomly selected patients before treatment for T1 to T4 oral tongue SCC were stained immunohistochemically with p21WAF1/CIP1. The percentages of positive nuclei that stained positive for tumor were determined.
Main Outcome Measures
The significance of prognostic parameters was tested by log-rank and Kaplan-Meier methods for the univariate analysis. The Cox proportional hazards regression model was used for multivariate analysis.
Expression of p21WAF1/CIP1 correlated inversely with T classification and clinical stage (P = .02 and P = .006, respectively) but not with N classification, tumor differentiation, or apoptosis-related variables such as p53 accumulation, apoptotic index, and Bax and Bcl-2 expression. Patients with tumors expressing high p21WAF1/CIP1 values had increased disease-specific survival time (P = .03). The correlation between p21WAF1/CIP1 expression and disease-specific survival was even more significant when restricted to p53-negative tumors (P = .002). A 2-parameter combination between p21WAF1/CIP1 expression and Bax expression or p21WAF1/CIP1 expression and p53 accumulation, respectively, revealed an enhanced prognostic potential (P<.001) when compared with single parameters.
The expression of p21WAF1/CIP1, particularly in combination with p53 accumulation or Bax expression, has prognostic value in oral tongue SCC.
TUMORIGENESIS and progression of malignant tumors are characterized by uncontrolled cell proliferation. In mammalians, the cell cycle progression is regulated by the cyclin-dependent kinases, which are activated by cyclin and inactivated by inhibitors of cyclin-dependent kinases including p21WAF1/CIP1. The p21 protein is the gene product of the WAF1, CIP1, or SD11 gene, which play an important role in regulating the G1-S transition of the cell cycle.1- 4 The biological significance of p21WAF1/CIP1 in controlling cell proliferation is not fully understood. However, in response to DNA damage, wild-type p53 accumulates and binds to the promoter region of the p21WAF1/CIP1 gene.2,5 This induces the expression of the p21WAF1/CIP1 that inhibits the activity of the cyclin/cyclin-dependent kinase complex to block cell cycle progression.3,5 In addition, p21WAF1/CIP1 may block DNA synthesis by direct binding to the proliferating cell nuclear antigen that is important in DNA replication.6,7 The levels of p21WAF1 may also be elevated by non–p53-dependent mechanisms.6,8 Furthermore, p21WAF1/CIP1 is associated with terminal differentiation and cell senescence, and thus may play a role in cell maturation and cell death.4,9
Results presented by previous studies regarding the correlation between expression of p21WAF1/CIP1 and prognosis are controversial. In several tumor types, expression or overexpression of p21WAF1/CIP1 seems to be associated with favorable clinical outcome.10- 12 Such correlation has, however, not been confirmed by others.13,14 These conflicting findings have also been reported in head and neck cancers.15- 18 So far, the prognostic significance of p21WAF1/CIP1 expression has scarcely been reported in tongue squamous cell carcinoma (SCC).
In the present study, we investigate the expression of p21WAF1/CIP1 in a series of oral tongue SCCs and the possible correlation between this parameter and clinical, histopathologic, p53, and apoptosis-related markers such as apoptotic index (AI) and Bax and Bcl-2 expression.19 Furthermore we have also tested whether a combination between parameters might have predictive value.
We analyzed formalin-fixed, paraffin-embedded pretreatment specimens from 80 randomly selected patients with oral tongue SCCs admitted to the Department of Otolaryngology, National Hospital, University of Oslo, Oslo, Norway, between 1983 and 1992. There were 34 women and 46 men ranging in age from 31 to 88 years (mean age, 63 years). All the tumors were classified according to the International Union Against Cancer (UICC) TNM classification of 1987. There were 51 T1 to T2 tumors and 29 T3 to T4 tumors. Thirty-seven patients had metastasis (N1, 22; N2, 14; and N3, 1). Clinical findings, treatment, and follow-up have been recorded prospectively. None of the patients were lost to follow-up. The range of follow-up is 0.1 to 11.6 years. Of 14 patients who died from diseases other than tongue SCC, only 3 patients had an observation time shorter than 3 years (1.7, 2.3, and 2.8 years, respectively). Thirty-seven patients died of the disease (15 T1-T2 and 22 T3-T4).
Five patients with small primary tumors (T1) were treated with surgery alone. Forty-three patients received combined treatment, 24 with radiotherapy (60-70 Gy) followed by surgery, and 19 with surgery followed by radiotherapy (50-60 Gy). Thirty-two patients with nonresectable tumors or who were unsuited for or refused surgery received radiotherapy alone (70-80 Gy). The dose used for elective radiotherapy of the neck was 50 Gy, whereas patients with neck metastases received 60 Gy followed by surgery whenever possible. Of 41 treatment failures, 4 patients with local recurrence underwent successful salvage procedures (all had T1-T2 tumors).
Serial 5-µm-thick sections were cut from the tissue blocks and mounted on gelatin-coated slides. One section was stained with hematoxylin-eosin for verification of the original histopathologic diagnosis and histologic differentiation. The histologic differentiation was graded by one of us (O.P.F.C.) as 1, good; 2, moderate; 3, poor; and 4, undifferentiated.
In brief, after being dewaxed in xylene, the sections were dehydrated in ethanol and rinsed in distilled water. For antigen retrieval, the slides were then incubated 5 times in 10 mM citric acid buffer (pH, 6.3) in a microwave oven (750 W) for 5 minutes. After cooling for 30 minutes at room temperature, the sections were washed in Tris buffer for 5 minutes and incubated overnight at 4°C with 1:100 dilution monoclonal antibody against p21 (WAF, Ab-1; Oncogene Science, Cambridge, Mass). The sections were incubated with biotinylated secondary antibody (1:500), washed with phosphate-buffered saline, incubated for 40 minutes with avidin-biotinylated peroxidase complex solution, then placed in a development solution containing 0.06% diaminobenzidine and 0.1% volume per volume hydrogen peroxide. The sections were finally counterstained with hematoxylin and mounted. For each batch of immunostained sections, 1 positive and 1 negative control (without the primary antibody) were included.
Ten systematically chosen fields were evaluated by means of an ×100 magnification oil-immersion lens. The percentages of tumor cells with, respectively, nuclear staining and cytoplasmic staining were determined. Specimens with fewer than 5% positive cells were considered negative. All percentages were determined by one of us (X.X.), and the results were based on the evaluation of at least 700 tumor cells in each section. Areas with pronounced inflammation, necrosis, or artificial damage were avoided. The assessments were performed without knowledge of the clinical outcome. Six months after the completion of the evaluation, 16 previously examined sections were randomly selected and percentages of tumor cells redetermined to test the reproducibility of the method.
The data were stored and analyzed by means of SAS 6.12 software (SAS Institute, Cary, NC). The χ2 test was used for comparison of group means. Correlation between parameters was performed by the Pearson test. Kaplan-Meier plots and log-rank tests were used to visualize and evaluate the significance of clinical variables and p21WAF1/CIP1 expression in relation to disease-specific survival. The DISCRIM procedure that computes various discriminant functions for classifying observations into groups on the basis of quantitative variables was used to screen possible combinations between 2 variables in relation to disease-specific survival. A case was censored when death resulted from causes other than the original tumor and when there was no evidence of residual tumors or recurrences at the last follow-up consultation. Prognostically significant parameters by the univariate analysis were further analyzed by means of the Cox proportional hazards regression model. Statistical significance was set at P<.05. Reproducibility was tested by means of the least squares regression analysis.
In normal epithelium adjacent to the tumors, p21WAF1/CIP1 was regularly expressed in the suprabasal region with nuclear staining, whereas nuclei in the superficial and/or basal regions were only occasionally stained. Cytoplasmic staining was also seen in some cases. In tumor cells, the p21WAF1/CIP1 expression was mainly nuclear, although cytoplasmic staining was also found in many cases. Some of the tumor cells showed both nuclear and cytoplasmic staining. Seventy-four tumors (92%) showed variable nuclear expression of p21WAF1/CIP1, with a mean value of 32% (range, 0%-88%), among them 25 cases (33%) with additional cytoplasmic staining. Three tumors showed cytoplasmic staining only.
Of 47 cases of p53-positive tumors, 45 showed p21WAF1/CIP1 nuclear expression. Eighteen of them showed a sum of more than 100% tumor nuclei stained for p21WAF1/CIP1 and p53 proteins, which indicates that parts of the tumor cells were positive for both p21WAF1/CIP1 and p53.
Results from p21WAF1/CIP1 nuclear expression were compared with histopathologic and clinical parameters and with our previously investigated apoptosis-related markers, including Bax and Bcl-2 expression, AI, and p53 accumulation.19 Expression of p21WAF1/CIP1 correlated inversely with T classification and clinical stage (P = .02 and P = .006, respectively). We found no correlation between p21WAF1/CIP1 and tumor differentiation, N classification, or any of the apoptosis-related parameters.
No correlation was found between p21WAF1/CIP1 nuclear expression and disease-specific survival when the overall mean value was used as the cutoff level. However, when using a plot diagram, we found that 23% of p21WAF1/CIP1 expression gave the best cutoff level regarding patient clinical outcome. Tumors with p21WAF1/CIP1 nuclear expression exceeding this 23% level had a significantly longer disease-specific survival time than those with values below 23% (P = .03; Figure 1). In all 33 cases of p53-negative tumors, p21WAF1/CIP1 nuclear expression correlated strongly with disease-specific survival using 23% as the cutoff level (P = .002; Figure 2). We found no correlation between p21WAF1/CIP1 cytoplasmic staining and patient clinical outcome.
Table 1 summarizes the findings of the log-rank analyses for clinical parameters, tumor differentiation, and p21WAF1/CIP1 expression. T and N classifications, clinical stage, treatment, and p21WAF1/CIP1 expression were significantly associated with disease-specific survival. Figure 3 shows the Kaplan-Meier plots for p21WAF1/CIP1 expression in relation to disease-specific survival (P = .04). Multiple regression analysis (Cox method) revealed clinical stage as the only significant independent variable.
In an attempt to elucidate a possible synergistic prognostic effect, we tested whether combination between p21WAF1/CIP1 expression and apoptosis-related parameters such as p53 accumulation, AI, and Bax and Bcl-2 expression might improve the prognostic ability with respect to disease-specific survival. The DISCRIM analysis indicated that there was an association between p21WAF1/CIP1 expression and p53 accumulation or Bax expression in discriminating between patients with favorable and unfavorable clinical outcome. We therefore selected 3 to 4 cutoff levels to obtain scores for each of the parameters evaluated according to their distribution in relation to disease-specific survival (Table 2). Since tumors with high values for p21WAF1/CIP1 and Bax expression correlated positively with favorable outcome, whereas high levels of p53 tended to correlate with poor prognosis, the scores of p21WAF1/CIP1 expression were added to the scores of Bax expression, but subtracted from the scores of p53 accumulation to test the prognostic significance of the new combined parameters.
Table 3 summarizes the findings of the log-rank analysis for individual parameters and the new combined parameters in relation to disease-specific survival and indicates that combined parameters seemed to be stronger prognostic indicators than any of the parameters alone. Figure 4 and Figure 5 show the cooperative effect between p21WAF1/CIP1 expression and, respectively, p53 accumulation and Bax expression in discriminating between survivors and nonsurvivors. Least squares regression analysis resulted in the R2 value of 0.8 for p21WAF1/CIP1 expression.
No correlation was found between the expression of p21WAF1/CIP1 and p53 accumulation. This is in agreement with the findings presented by 2 other studies in head and neck SCC.16,18 Previous studies have suggested that p21WAF1/CIP1 expression is induced by wild-type p53.2,5 This was supported by 2 studies in breast and ovarian cancers where p21WAF1/CIP1 expression was inversely correlated with p53 accumulation,12,20 but not confirmed in 2 other studies of breast carcinoma and epithelial ovarian carcinomas.13,21 Consequently, it is possible that p53-dependent and p53-independent mechanisms may be involved in p21WAF1/CIP1 expression.22,23 This is supported by the findings of similar frequencies of p21WAF1/CIP1 expression in p53 wild-type tumors and p53 mutant tumors in head and neck SCC.16,18,22 A study on oral premalignant and malignant lesions also suggested that expression of p21WAF1/CIP1 is induced by p53-dependent as well as p53-independent mechanisms during oral tumorigenesis.22 In the present study, nearly all the p53-positive tumors expressed p21WAF1/CIP1. Eighteen tumors showed a sum of more than 100% tumor nuclei stained for p21WAF1/CIP1 and p53 proteins, which indicates that parts of the tumor cells were positive for both p21WAF1/CIP1 and p53. These findings support the finding that p53-dependent and p53-independent mechanisms are involved in p21WAF1/CIP1 expression in tongue SCC.
In the present study, we show that patients with tumors that have high p21WAF1/CIP1 expression values have a more favorable clinical outcome than those with low values. This is in agreement with the findings from studies of several other types of tumor such as breast carcinoma, lung SCC, epithelial ovarian cancer, and laryngeal carcinoma.10- 12,15 Contradictory results have also been reported in breast carcinoma, superficial bladder tumors, and head and neck cancer,13,14,16 whereas no correlation was found between p21WAF1/CIP1 expression and prognosis in laryngeal SCC.17 These contradictory results for head and neck cancer as well as for tumors at other sites indicate that the association between prognosis and expression of p21WAF1/CIP1 is complex and may vary from tumor type to tumor type. In addition, we clearly show that the impact of p21WAF1/CIP1 expression on prognosis is stronger among p53-negative tumors than among those with p53 accumulation. This suggests that the p53-dependent regulation of p21 is more important for disease-specific survival than the p53-independent pathway.
Our findings indicate that the evaluation of the combined variable p21WAF1/CIP1 expression and p53 accumulation may have more prognostic value discriminating patients with favorable and unfavorable clinical outcome in tongue SCC than evaluation of single variables. This observation is supported by studies on breast carcinoma and invasive ductal carcinoma of the pancreas.13,23 Unlike the method used by others who compared different groups of patients using 1 cutoff level (positive/negative) for p21WAF1/CIP1 and p53, we selected several cutoff levels to obtain scores for p21WAF1/CIP1 and p53 accumulation according to their distribution with regard to the clinical outcome. With this method we demonstrated a more complex and synergistic relationship between these 2 variables in discriminating patients with good and poor prognosis. Interestingly, p53 in itself was not a statistically significant prognostic variable in this series of patients. However, when p53 was combined with p21WAF1/CIP1, this new variable seemed to be a strong prognostic parameter.
Concerning the relationship between p21WAF1/CIP1 expression and apoptosis-related markers including AI and Bax and Bcl-2 expression, we found no correlation. However, our results showed that evaluation of p21WAF1/CIP1 and Bax expression combined improves prognostic ability over single parameters alone. This finding is in agreement with our previous observation in head and neck carcinomas where evaluation of the combination of proliferative and apoptotic markers was of greater prognostic value than evaluation of individual variables.24,25 This observation may suggest that such a combination of p21WAF1/CIP1 and Bax expression reflects different aspects of cell proliferation and apoptosis.
Our investigation suggests that p21WAF1/CIP1 expression has prognostic value in SCC of the tongue. Evaluation of p21WAF1/CIP1 expression in combination with, respectively, p53 accumulation and Bax expression provides considerably better prognostic information than evaluation of any individual variables alone.
Accepted for publication January 17, 2002.
This study was supported by funding from the Norwegian Cancer Society, Oslo.
We thank the Norwegian Cancer Society for generously supporting this study and Aasa Schjölberg for excellent technical assistance.
Corresponding author and reprints: Xin Xie, MD, PhD, Department of Otolaryngology, The National Hospital, 0027 Oslo, Norway (e-mail: email@example.com).