P21^Cip1/WAF1 expression is strongly associated with HPV-positive tonsillar carcinoma and a favorable prognosis

Abstract

Human papillomavirus is involved in the carcinogenesis of tonsillar squamous cell carcinomas. Here, we investigated the expression and the prognostic value of key cell cycle proteins in the pRb and p53 pathways in both human papillomavirus type 16-positive and -negative tonsillar squamous cell carcinomas. Using immunohistochemistry, 77 tonsillar squamous cell carcinomas with known human papillomavirus type 16 status and clinical outcome were analyzed for expression of Ki67, p16^INK4A, cyclin D1, pRb, p14^ARF, MDM2, p53, p21^Cip1/WAF1, and p27^KIP1. Results were correlated with each other and with clinical and demographic patient data. A total of 35% of tonsillar carcinomas harbored integrated human papillomavirus type 16 DNA and p16^INK4A overexpression, both being considered essential features for human papillomavirus association. These tumors also showed the overexpression of p14^ARF (P<0.0001) and p21^Cip1/WAF1 (P=0.001), and downregulation of pRb (P<0.0001) and cyclin D1 (P=0.027) compared with the human papillomavirus-negative cases. Univariate Cox regression analyses revealed a favorable survival rate for non-smokers (P=0.006), as well as for patients with T1-2 tumors (P<0.0001) or tumors showing low expression of cyclin D1 (P=0.028), presence of human papillomavirus and overexpression of p16^INK4A (P=0.01), p14^ARF (P=0.02) or p21^Cip1/WAF1 (P=0.004). In multivariate regression analyses, smoking and tumor size, as well as expression of cyclin D1 and p21^Cip1/WAF1, were found to be independent prognostic markers. We conclude that human papillomavirus positivity in tonsillar squamous cell carcinomas strongly correlates with p21^Cip1/WAF1 and p14^ARF overexpression and downregulation of pRb and cyclin D1. In particular p21^Cip1/WAF1 overexpression is an excellent favorable prognosticator in tonsillar squamous cell carcinomas.

Main

Head-and-neck squamous cell carcinoma is the sixth most prevalent malignancy in the world, contributing 6% of new cancer cases annually worldwide.^{1, 2} These tumors have a 5-year survival rate of approximately 50%, which has not improved in the last two decades.³ Well-recognized risk factors in the etiology of head-and-neck squamous cell carcinomas are extensive tobacco and alcohol consumption in ∼90% of cases, as well as oncogenic human papillomaviruses (HPVs), predominantly HPV type 16.^{3, 4} Interestingly, the association of HPV is strongest for tonsillar squamous cell carcinoma with a prevalence up to 50%.^{5, 6, 7, 8} The diagnosis of HPV positivity in head-and-neck squamous cell carcinomas appears to have significant prognostic implications. In one study, these patients had <50% chance of dying from the disease compared with HPV-negative tumors.⁹

It has been shown that there are several differences between HPV-positive and -negative head-and-neck squamous cell carcinomas. Despite the fact that HPV positivity in head-and-neck squamous cell carcinomas is an indicator for favorable prognosis, from a clinical point of view these tumors are often poorly differentiated^{4, 6, 10, 11, 12} and metastasized to lymph nodes at presentation.^{10, 11} Furthermore, HPV-positive tumors are often smaller at first diagnosis (diameter ≤4 cm),¹³ and associated with low/no exposure to alcohol and tobacco.^{10, 11} At the molecular level, the functional inactivation of two key tumor suppressor proteins, ie, p53 and pRb by the HPV-derived oncoproteins E6 and E7, often result in the downregulation of p53, pRb, cyclin D1, and a strong upregulation of p16^INK4A in HPV-positive tumors.^{5, 7, 14, 15, 16} HPV-negative tumors, in contrast, often show inactivation of p16^INK4A, p53 overexpression as a result of gene mutations, cyclin D1 gene amplification and overexpression, as well as EGFR accumulation.^{3, 17, 18, 19}

The literature, however, shows conflicting data with respect to HPV-associated characteristics and clinical outcome of head-and-neck squamous cell carcinomas. First, although many studies describe a significant association between HPV presence and favorable prognosis, some studies did not find such a correlation.^{20, 21, 22} Second, HPV has also been identified in head-and-neck squamous cell carcinomas of smokers, significantly reducing its favorable effect on clinical outcome.¹³ Furthermore, despite the fact that regional lymph node metastasis is considered as the most important prognostic factor in head-and-neck squamous cell carcinomas,²³ this parameter seems to be unreliable in tonsillar squamous cell carcinomas.^{24, 25} Finally, some studies reported overexpression and/or p53 mutations almost exclusively in HPV-negative tumors,²⁰ whereas others have found that HPV infection and p53 alterations can coexist.²⁶ This may have a strong effect on survival, because it has been indicated that tumors with intact p53 are still capable of inducing apoptosis in response to radiation therapy, which results in a favorable clinical outcome.²⁷

This study was undertaken to investigate the expression of key cell cycle proteins in the pRb pathway (p16^INK4A cyclin D1, p27^Kip1 pRb) and the p53 cascade (p14^ARF MDM2, p53 and p21^Cip1/WAF1), using a series of 77 tonsillar squamous cell carcinomas for which the HPV16 status and the clinical follow-up data were available. Tonsillar squamous cell carcinomas show the highest prevalence of oncogenic HPV and are thus ideally suited to search for molecular and clinicopathological differences induced by either HPV, or tobacco and alcohol consumption.

Materials and methods

Tumor Material and Patient Data

Formaldehyde-fixed, paraffin-embedded archival biopsy and resection material of primary tonsillar squamous cell carcinomas from 77 patients was selected from the archives of the Department of Pathology, Maastricht University Medical Center, The Netherlands. This material had been previously examined for HPV-specific DNA by means of PCR and FISH, and only showed the presence of oncogenic HPV16.¹³ Demographic data, including age at diagnosis, gender, alcohol and tobacco exposure, treatment modality, and date and cause of death were obtained from the medical records. Tumor site, degree of differentiation (ie, well, moderately, or poorly differentiated), and TNM classification were determined from review of pathological, radiological, and surgical reports. All patients were treated by surgery, radiotherapy, chemotherapy, or a combination, irrespective of HPV status. The study protocol was approved by the institutional ethical committee, and all patients gave informed consent. Table 1 provides demographic and clinical features of the 77 patients included in this study. Fifty-seven patients were male and 20 patients were female. The mean age at diagnosis was 58.8 (range 39–87) years. Data on smoking and alcohol intake were obtained from 76 and 75 patients, respectively. A total of 64 (84%) of 76 patients were smokers (≥1 cigarette, pipe, and/or cigar per day) and 46 (61%) of 75 patients were classified as drinkers (consumption of >2 whiskey equivalents per day (one whiskey equivalent is ∼10 g alcohol)). Forty-two (55%) patients used both tobacco and alcohol, whereas only eight (10%) patients had not been subjected to these intoxications. Thirty-nine (51%) patients had a tumor with a diameter ≥4 cm, and 55 (71%) had lymph node metastasis at the time of diagnosis. Tumor grade was poor or moderate in 63 (82%) patients, well differentiated in 10 (13%) patients, unavailable in 3 patients (4%), and 1 patient (1%) had a carcinoma in situ. Following primary treatment, 29 (38%) patients never became disease-free, 16 (21%) developed a recurrent disease (locoregional, regional, or distant), and 32 (42%) patients remained disease-free after primary treatment.

Table 1 Clinicopathological features and cell cycle protein expression in relation to HPV status

A series of 4-μm thick sections was cut from the specimens for hematoxylin–eosin staining and a detailed histopathological classification (FJB) was given according to the criteria of the World Health Organization.²⁸ Furthermore, we applied immunohistochemistry to visualize Ki67, p16^INK4A cyclin D1, pRb, p14^ARF MDM2, p53, p21^Cip1/WAF1, and p27^Kip1 expression.

Immunohistochemical Staining

Immunohistochemical protein staining on 4-μm thick formaldehyde-fixed, paraffin-embedded tissue sections was carried out as described earlier.¹⁵ Briefly, sections were deparaffinized and subsequently pretreated with 2% H₂O₂ in methanol for 30 min to quench endogenous peroxidase activity. Antigen retrieval was carried out by microwave heating in 0.01-M citrate buffer (pH 6.0). The primary antibodies used to detect Ki67, p16^INK4A cyclin D1, pRb, p14^ARF MDM2, p53, p21^Cip1/WAF1, and p27^Kip1 are listed in Table 2. After incubation with a biotinylated secondary antibody, immunohistochemical detection was performed by an avidin–biotinylated peroxidase complex (ABC) procedure (Vectastain-Elite-ABC kit; Vector, Burlingame, USA). Peroxidase activity was detected using 0.5 mg/ml diaminobenzidine/2% H₂O₂. Sections were counterstained with hematoxylin and mounted in Entellan (Merck, Darmstadt, Germany). In each analysis, negative and positive controls were included. Analysis was carried out by three independent observers (JJM, EJMS, and SMHC), and in case of interobserver variations, consensus was reached by combined examination of the slides. Both staining intensity (−, +/−, +, ++, +++) and the percentage of stained tumor cells were scored. Evaluation criteria for positive scoring of each of the cell cycle proteins are listed in Table 2.

Table 2 Primary antibody characteristics and used evaluation criteria

Statistical Analysis

The study population consisted of 77 patients with a tonsillar squamous cell carcinoma diagnosed between 1992 and 2001. Tumors were considered to be HPV-associated if they showed HPV16 presence by in situ hybridization analysis, in addition to the overexpression of p16^INK4A, as detected by immunohistochemistry. Factors associated with HPV status were selected on cross-tabulations, which were analyzed by the use of the two-tailed Fisher's exact test and/or χ²-test. The maximum significance levels are indicated for all analyses (P≤0.05). Disease-specific survival curves were calculated using the Kaplan–Meier method. Survival was calculated from the date of diagnosis until patient's death or until the last date the patient was known to be alive (this ranged from 16 to 141 months). Patients who died of other causes than tonsillar carcinoma were considered censored observations in the disease-specific survival analyses. Disease-free survival was calculated from the date of diagnosis until the date of recurrence (local, regional, or distant, whichever occurred first). Patients without recurrence were censored at the date of the last follow-up or at the date of death. The statistical significance of differences between survival times as determined by the log-rank test in univariate analysis. Multivariate analyses were performed using the Cox proportional hazards model. Variables in the multivariate model included HPV association, T classification, smoking, and cyclin D1-, p14^ARF- and p21^Cip1/WAF1 expression. Variables remained in the model if their P-values were <0.10. All calculations were performed by the use of the SPSS Base System version 12.0.1.

Results

Cell Cycle Protein Expression

Seventy-seven tonsillar squamous cell carcinomas were examined for expression of the cell cycle proteins Ki67, p16^INK4A cyclin D1, pRb, p14^ARF MDM2, p53, p21^Cip1/WAF1, and p27^Kip1, using the primary antibodies and evaluation criteria stated in Table 2. Representative immunostaining results in normal epithelium, and tumor tissues are shown in Figure 1. The frequency of tumors exhibiting the expression of the respective proteins is presented in Figure 2a. In all, 75 out of 77 (97%) tonsillar squamous cell carcinomas showed strong, nuclear Ki67 expression in 30–90% of the tumor cells (Figure 1a). Also nuclear pRb staining was often observed in tonsillar carcinomas (62 out of 77 cases, 81%), although two subsets were identified with a difference in staining intensity (ie, tumors showing < or ≥ intense immunostaining than present in the adjacent squamous epithelium. Figure 1d and j, respectively). A significant association was seen between the overexpression of pRb (a higher nuclear staining intensity in tumor cells than in the adjacent normal squamous epithelium) and p53 accumulation (P=0.017). Both overexpression of p53 and p21^Cip1/WAF1 correlated independently with the accumulation of MDM2 (Figure 1k; P=0.030 and 0.004, respectively), but not with each other (data not shown; Supplementary Table 1). In addition, there was a significant association between the overexpression of p16^INK4A and that of both p14^ARF (P<0.0001) and p21^Cip1/WAF1 (P=0.027; Figure 1b, c and f). In case of p14^ARF overexpression in the tumor, a diffuse nuclear staining pattern was more often observed than nucleolar staining (Figure 1c, above and below, respectively). Furthermore, p16^INK4A accumulation strongly correlated with both downregulation of pRb (P<0.0001) and cyclin D1 (P=0.035; Figure 1d and e).

Figure 1

Representative examples of immunohistochemical staining on tissue sections for (a) Ki67, (b) p16^INK4A (c) p14^ARF (d, j) pRb, (e, i) cyclin D1, (f) p21^Cip1/WAF1 (g, h) p53, (k) MDM2, and (l) p27^Kip1. Sections are from HPV-positive tonsillar squamous cell carcinomas (a–g), HPV-negative tonsillar squamous cell carcinomas (h–l) and adjacent squamous epithelium (a, b, d–g; lower image). Evaluation criteria as described in Materials and methods section. (a) Strong nuclear Ki67 staining in tumor cells and parabasal epithelial cells; (b) strong and diffuse p16^INK4A staining in tumor cells, predominantly cytoplasmatic; epithelium negative; (c) strong and diffuse nuclear p14^ARF staining (higher image). Some tumors show a rather nucleolar-like immunostaining (lower image). (d) Weak/no nuclear pRb staining in tumor cells in comparison with adjacent normal squamous epithelium. (e) Weak/no nuclear cyclin D1 staining in tumors cells; parabasal epithelial cells weakly positive; (f) strong nuclear p21^Cip1/WAF1 staining in tumor cells; some (para)basal epithelial cells weakly positive; (g) low frequency of tumor cells positive for nuclear p53; some (para)basal epithelial cells weakly positive; (h) high frequency of tumor cells positive for nuclear p53; (i) high frequency of tumor cells positive for nuclear cyclin D1; (j) strong nuclear pRb staining in tumor cells; (k) low frequency of tumor cells showing strong and nuclear MDM2 staining; (l) weak to strong nuclear immunostaining for p27^Kip1 in tumor cell areas as well as in adjacent lymphocyte areas. Magnification: × 40 (a–c (higher image), d–i, k–l), × 120 (j) and × 200 (c; lower image).

Figure 2

(a) Cell cycle protein expression in tonsillar squamous cell carcinomas according to the evaluation criteria presented in Table 2. (b) Fraction of HPV-positive and HPV-negative tonsillar squamous cell carcinomas showing expression of individual cell cycle markers (see Table 2 for the used criteria for positivity of each individual cell cycle marker).

Staining of normal tonsillar squamous epithelium was seen in the basal or parabasal cell layers for Ki67, cyclin D1, pRb, p21^Cip1/WAF1, and p53, varying from low to high intensities (Figure 1a, b, d–g). This squamous epithelium, present in the individual sections, was used as an internal positive control. In p14^ARF- and p27^Kip1-stained sections, adjacent lymphocyte infiltration often showed a strong cytoplasmatic and nuclear staining, respectively, and therefore served as the internal positive control (Figure 1l).

Correlations Between HPV Status, Expression of Cell Cycle Proteins, and Clinicopathological Variables

The correlations between each cell cycle marker and HPV status are shown in Table 1 and Figure 2b. HPV-associated tonsillar squamous cell carcinomas showed significantly more often overexpression of p14^ARF (P<0.0001) and p21^Cip1/WAF1 (P=0.008), and downregulation of pRb (P<0.0001) and cyclin D1 (P=0.027) than HPV-negative tumors. P53 accumulation tended to be associated with the absence of HPV (P=0.079).

The male/female ratio and the age distribution were similar for the HPV-positive and HPV-negative subgroups. Smoking and alcohol abuse were seen significantly more often in the HPV-negative patient group (P<0.0001 and P=0.024, respectively). HPV-positive tumors tended to be smaller than HPV-negative tumors (P=0.065). Male patients were more often diagnosed with a T3–4 tumor (P=0.006), and were more often smokers (P=0.011).

Indicators for Disease-Specific Patient Survival

To determine whether or not cell cycle protein expression patterns and clinicopathological parameters can be used as indicators of prognosis, we correlated these with the disease-specific survival data of patients with tonsillar squamous cell carcinoma (Table 3). Two patients died postoperatively, due to bleeding and aspiration, and from one patient, no follow-up data were available. These patients were excluded from the analyses. Follow-up time ranged from 0 to 141 months, with a mean of 30 months. A total of 41 (55%) of 74 patients died as a consequence of tonsillar carcinoma. The survival after 5 years was 31% for patients with a HPV-negative tumor and 69% for patients with a HPV-positive carcinoma (Hazard ratio (HR)=0.4; 95% confidence interval (CI)=0.2–0.8). Besides the absence of HPV, the following parameters were also significantly associated with a shorter disease-specific survival according to Univariate Cox regression analysis: (1) smoking (HR=5.8; 95% CI=1.4–24.1), (2) a tumor diameter ≥ 4 cm (HR=3.1; 95% CI=1.6–6.0), (3) development of recurrent disease (HR=14.1; 95% CI=3.9–51.1), (4) no/low expression of either p14^ARF (HR=2.2; 95% CI=1.1–4.3) or p21^Cip1/WAF1 (HR=2.7; 95% CI=1.3–5.5), and (5) positive cyclin D1 immunostaining in >50% of tumor cells (HR=2.2; 95% CI=1.1–4.7). Gender, age at diagnosis, alcohol use, tumor grade, TNM-stage, lymph node status, and the remaining cell cycle markers were not related to disease-specific survival.

Table 3 Influence of HPV-related and clinicopathologic parameters on disease-specific survival in 74 patients with TSCC, as determined by univariate Cox proportional hazard regression analysis

The parameters that were significantly correlated with disease-specific survival in the univariate analysis, ie, HPV status, tumor size, smoking, and immunostaining of p14^ARF p21^Cip1/WAF1, and cyclin D1, were included in the multivariate Cox regression analysis. Development of recurrent disease was not included for multivariate analysis, because this factor cannot be predicted at time of diagnosis, so the clinical impact is of less importance. Table 4 shows that four of the six parameters, ie, tobacco consumption, tumor diameter ≥4 cm, no/low p21^Cip1/WAF1 immunostaining, or strong cyclin D1 immunostaining, were the most optimal indicators of cancer-specific death, with tumor size and p21^Cip1/WAF1 immunostaining being the most significantly correlated. In Figure 3, the Kaplan–Meier curves for these four parameters are shown.

Table 4 Multivariate analysis, according to Cox proportional hazard regression analysis, of the patient and tumor characteristics related to disease-specific mortality

Figure 3

Kaplan–Meier survival curves according to (a) tumor size, (b) smoking status, (c) p21^Cip1/WAF1 expression, and (d) strong cyclin D1 expression.

Discussion

In this study, we examined the expression of cell cycle-related constituents in a series of 77 tonsillar squamous cell carcinomas with the goal to determine their role in HPV-dependent and HPV-independent carcinogenesis. Furthermore, their prognostic value was evaluated by correlating the results with clinical follow-up data. Our results show that HPV16-positive tumors exhibit p14^ARF and p21^Cip1/WAF1 overexpression and downregulation of pRb and cyclin D1 in contrast to HPV16-negative tumors. Secondly, tumor size and p21^Cip1/WAF1 positivity are the strongest independent indicators for a favorable outcome in patients with tonsillar squamous cell carcinoma.

All tonsillar squamous cell carcinomas, except one HPV-positive and one HPV-negative tumor, showed a high expression of Ki67, indicating that almost all tumors contained a high percentage of proliferative cells, which is in agreement with other studies on head-and-neck squamous cell carcinomas.³⁷ In addition, high expression levels of inhibitors of apoptosis, such as BclX_L and survivin, have been reported in head-and-neck squamous cell carcinomas.^{38, 39, 40, 41} Expression of pRb was detected in 81% of the tumors. Approximately half of all tumors showed strong nuclear expression in the tumor cells, with expression levels equal to or higher than the adjacent normal squamous epithelium, whereas pRb downregulation was observed in the remaining cases. In our study, p53 overexpression was observed in little more than half of all tonsillar carcinomas, which is consistent with other studies.^{7, 42} Immunostaining of all other cell cycle proteins was evident in less than 50% of tumors.

We noticed that the tumors with the low expression levels of pRb showed the overexpression of p16^INK4A. Accumulation of this latter protein has been strongly associated with the presence of oncogenic HPV in oropharyngeal carcinomas,^{7, 10, 14, 15} which was also evident in the underlying study. These HPV16-positive tonsillar squamous cell carcinomas furthermore showed the downregulation of cyclin D1, next to the accumulation of p14^ARF and p21^Cip1/WAF1. In addition, p53 expression was less profound in these HPV-positive tumors.

In HPV-associated tumors, the oncoprotein E7 interacts with pRb, resulting in its degradation. As a result, p16^INK4A is upregulated and cyclin D1 is downregulated.⁴³ This is in agreement with Andl et al,⁵ who suggested that E7 might overcome the need for cyclin D1 in the G1 phase of the cell cycle, because it interacts with the cyclin D1-binding site on pRb.^{7, 44} Indeed, high expression levels of cyclin D1 were predominantly observed in HPV-negative tonsillar carcinomas, most probably pointing to cyclin D1 gene amplification in these cases.

The p14^ARF gene is a target for the transcription factor E2F, which promotes its expression particularly in the HPV-positive tumors, as has become evident in this study. In the HPV-negative tumors both p14^ARF as well as p16^INK4A were often undetectable, which is in accordance with the fact that their expression is downregulated in most head-and-neck squamous cell carcinomas due to gene inactivation at the 9p21 locus.^{45, 46} In normal tissue cells, p14^ARF upregulation might lead to p53 upregulation, but in HPV-positive tonsillar squamous cell carcinomas, this is usually counteracted by means of functional inactivation of wild-type p53 by the viral E6 protein. In contrast and as observed in our study, p53 is frequently upregulated in HPV-negative tumors due to mutations in the TP53 gene as a result of exposure to tobacco and/or alcohol.^{10, 16, 20, 27, 47}

Although p21^Cip1/WAF1 is known to be a downstream effector of p53,^{48, 49} it was surprising to find overexpression in HPV-positive tumors harboring low or no detectable levels of p53. Such observations have also been reported by Milde-Langosch et al⁵⁰ in HPV-associated uterine cervical tumors and suggest that also p53-independent mechanisms may lead to p21^Cip1/WAF1 accumulation as described previously.^{51, 52} Indeed, in HPV-positive cancer cells, p21^Cip1/WAF1 expression appears to be inducible,⁵³ although it remains unclear why under these conditions E7 cannot inactivate p21^Cip1/WAF1 by direct interaction.¹⁴

Despite the strong association of p21^Cip1/WAF1 overexpression with HPV positivity in most tonsillar squamous cell carcinomas, we found p21^Cip1/WAF1 protein accumulation also in some HPV-negative tumors, which is in accordance with a study of Li et al.⁵⁴ These authors examined 67 tonsillar squamous cell carcinomas for HPV involvement, but did not find p21^Cip1/WAF1 expression being associated with HPV positivity. An explanation for this paradox might be on the one hand the use of a different p21^Cip1/WAF1-specific primary antibody and different criteria to assess p21^Cip1/WAF1 upregulation (>20% of positive nuclei versus >10%) in comparison with our and other studies,^{29, 30, 35, 36} and on the other hand the use of only PCR to determine the presence of HPV. To assess a firm association between virus and tumor cells, namely, it is recommended to carry out additional tests, such as p16^INK4A immunostaining and/or FISH, which has been applied in our study.^{55, 56}

In the univariate and multivariate statistical analyses, we found that p21^Cip1/WAF1 overexpression was a highly significant indicator of favorable prognosis in tonsillar squamous cell carcinomas independent of HPV status. Expression of p21^Cip1/WAF1 has also been associated with a favorable survival in patients with tongue squamous cell carcinomas, and in patients with ovarian, superficial bladder, gastric, colorectal, and esophageal cancers.^{29, 43, 57, 58} In one other study on tonsillar squamous cell carcinomas and two studies on laryngeal squamous carcinomas, no correlation was reported,^{36, 54, 59} and in one study on head-and-neck squamous cell carcinomas derived from multiple head-and-neck localizations even a negative correlation with survival was described,¹⁹ which may be explained by the heterogeneous tumor population. Other strong indicators for a favorable prognosis in our study include tumor diameter <4 cm, low/no expression of cyclin D1, and low/no tobacco consumption. In accordance with most other studies on HPV-related oropharyngeal carcinomas, we also found a favorable disease-specific survival in patients with HPV16-positive tonsillar squamous cell carcinomas,^{12, 46, 60, 61, 62, 63, 64} and p14^ARF overexpression,³¹ although with less significance than the indicators described above. We noticed that lymph node status, which is generally considered as the most important prognostic factor in head-and-neck squamous cell carcinomas,²³ had little value in our series of tonsillar squamous cell carcinomas, which is in concordance with other studies.^{24, 25}

In summary, we can conclude that HPV16-positive tonsillar squamous cell carcinomas exhibit overexpression of p14^ARF and p21^Cip1/WAF1 as well as downregulation of pRb and cyclin D1, and that strong immunostaining for p21^Cip1/WAF1 appears to be one of the most potent indicators for favorable prognosis in these tumors.