Methods

TISSUE SELECTION AND STANDARD HISTOLOGY

A highly selective cohort of 38 patients (30 men, eight women; age range, 17–76 years; median, 49) was investigated. Archived surgical specimens were retrieved from the files of the department of pathology, Military Hospital Ulm, Ulm/Donau, Germany. Material with exact clinical documentation of the origin and extension of the lesions was used exclusively. Furthermore, only specimens with an adequate amount of lesional material were investigated. Cases showing insufficient fixation, necrosis, haemorrhage, and/or atrophy were excluded. Histological examination for adequacy of lesional tissue and classification of specimens was carried out by two independent observers using haematoxylin and eosin stained slides. Differences between their individual reports were resolved by re-examination and consensus. Non-papillomatous nasal mucosa was studied in seven cases with either normal nasal mucosa or minimal chronic inflammation. Specimens with more than minimal chronic inflammation, active inflammation, dysplasia, or carcinoma were excluded. Twenty six endophytic sinonasal papillomas were studied, comprising 20 inverted papillomas and six cylindrical cell papillomas. Specimens with dysplasia or associated malignancies were excluded from this group. Five squamous cell carcinomas associated with endophytic sinonasal papillomas were investigated.

IMMUNOHISTOCHEMISTRY

Additional 5 μm paraffin wax embedded sections were prepared and immunohistochemistry was carried out using the streptavidin–biotin–peroxidase technique, as reported previously.¹⁶ For all antibodies, a microwave based antigen retrieval protocol was used for a total of 30 minutes in citrate buffer (pH 6.0).¹⁷ The NCL-Waf 1 antibody (Novocastra, Newcastle upon Tyne, UK) was used for p21^waf1/cip1 staining, at a dilution of 1/20. Gastric body mucosa samples provided positive controls. Distinct nuclear staining was found in foveolar cells on the gastric tips, in addition to terminally differentiated cells in the glands. Furthermore, reproducible strong staining in vascular smooth muscle cells was seen in all specimens, which served as an internal positive control. The presence of human papillomavirus (HPV) infection was investigated using the NCL-PVp antibody (Novocastra), at a 1/100 dilution. This antibody detects all HPV types in paraffin wax embedded tissues.¹⁸ Positive controls were provided by a carcinoma in situ from the ectocervix. Strong nuclear and perinuclear staining was seen in most of the tumour cells. Furthermore, the expression of the p53 protein was studied using the antibody DO-7 (Dako, Hamburg, Germany), at a 1/50 dilution. Positive controls were carried out using a moderately differentiated gastric adenocarcinoma previously shown to be p53 positive. Intensive nuclear staining was found in more than 90% of cells on the invasion front. For all antibodies, negative controls were provided using normal horse serum with the same dilution.

EVALUATION OF STAINING RESULTS

Assessment of immunohistochemistry was done independent from classification of specimens. The following semiquantitative p21^waf1/cip1 expression scores were used: negative staining, less than 5% positive nuclei, 5–15% positive nuclei, 16–25% positive nuclei, and more than 25% positive nuclei. Basal/parabasal cells, medial cell layers, and surface cells of the epithelium were evaluated. In non-papillomatous nasal mucosa and cylindrical cell papillomas columnar epithelium were assessed. In inverted papillomas, columnar, transitional, and squamous epithelium were evaluated. In squamous cell carcinomas, representative regions on the invasion front of the tumour were investigated. Staining results were compared between these groups. In addition, HPV infection and p53 protein overexpression were studied and correlated with p21^waf1/cip1 expression. Physiological expression of the p53 protein was defined as less than 15% immunoreactive basal/parabasal cells. Overexpression of the p53 protein was defined as positive staining of more than 15% basal/parabasal cells, and/or immunoreactivity more superficial than the parabasal cell layer. Furthermore, overexpression of the p53 protein was subgraded in cases showing more or less than 40% immunoreactive nuclei.^19,20

Results

NON-PAPILLOMATOUS NASAL MUCOSA

All specimens were negative for p21^waf1/cip1 staining. HPV infection and p53 protein overexpression were not found.

EXPRESSION OF P21^waf1/cip1IN INVERTED PAPILLOMAS

p21^waf1/cip1 was expressed in 16 of 20 cases (table 1). Immunoreactivity for p21^waf1/cip1 was restricted to surface cells in five specimens (fig 1), but involved basal/parabasal cells in 11 specimens. Notably, in all cases immunoreactivity in basal/parabasal cells was separated from positive surface cells by negative medial cell layers of the epithelium. In some cases, positive staining in surface cells only along with positive staining in basal/parabasal cells only was detected in different regions within the same tissue section (fig 2). Frequently, the proportions of p21^waf1/cip1 positive basal/parabasal cells were enhanced in transitional and squamous epithelium compared with columnar epithelium. Positive staining for p21^waf1/cip1 in basal/parabasal cells colocalised with p53 protein overexpression (figs 3 and 4). The expression of p53 protein predominantly involved basal/parabasal cells of the epithelium. In all inverted papillomas, less than 40% of the cells were immunoreactive for p53 protein.

EXPRESSION OF P21^waf1/cip1IN CYLINDRICAL CELL PAPILLOMAS

All specimens showed p21 ^waf1/cip1 expression only in surface cells (table 2; fig 5). HPV infection and p53 protein overexpression were present in all cases.

EXPRESSION OF P21^waf1/cip1 IN SQUAMOUS CELL CARCINOMAS

One of five squamous cell carcinomas showed p21^waf1/cip1 expression (fig 6). HPV infection was found in two cases and p53 protein overexpression was detected in all tumours.

Discussion

p21^waf1/cip1 was not expressed in non-papillomatous nasal mucosa. Currently, it is thought that tissue specific p21^waf1/cip1 expression occurs in terminally differentiated epithelial cells.^6,21,22 For instance, normal breast epithelium shows p21^waf1/cip1 expression in less than 1% of cells.²³ In contrast, normal colonic epithelium shows strong p21^waf1/cip1 expression in the upper third of the crypts.²⁴

We found p21^waf1/cip1 expression in 16 of 20 inverted papillomas. Positive staining was seen in surface cells and basal/parabasal cells of the epithelium. Immunoreactivity for p21^waf1/cip1 in surface cells was independent of p53 protein overexpression. Hence, p53 independent activation of p21^waf1/cip1 expression along with terminal differentiation of surface cells in inverted papillomas can be postulated. A similar concept has recently been reported from oral squamous epithelium, in which p53 independent p21^waf1/cip1 expression involves the spinous cell layer undergoing terminal differentiation.^10,11,25

The expression of p21^waf1/cip1 in basal/parabasal cells in inverted papillomas colocalised with p53 protein overexpression. At least two wild-type p53 protein responsive elements exist in the promoter region of the waf1/cip1 gene.²⁶ Expression of p21^waf1/cip1 mediates the antiproliferative effect of p53 protein by the suppression of DNA replication and G₁ cell cycle arrest.^{3,4,7,26–28} Hence, it can be hypothesised that wild-type p53 dependent p21^waf1/cip1 expression occurs in basal/parabasal cells in inverted papillomas. However, the antibody DO-7 used in our study recognises both wild-type and mutant p53 protein.²⁹ The expression of p53 protein in less than 40% of cells was consistently found in our inverted papillomas. Recent studies have associated p53 protein expression in less than 40% of nuclei with the accumulation of wild-type rather than mutant p53 protein.^30–32 Furthermore, a strong association between wild-type p53 protein and non-dysplastic inverted papillomas, in contrast to mutant p53 protein expression found in dysplastic specimens, has recently been reported.³³ Dysplastic inverted papillomas were excluded from investigation in this study. Hence, it is probable that wild-type p53 protein was expressed in our exclusively non-dysplastic inverted papillomas, although further investigations involving molecular analysis of the p53 gene status are necessary to confirm these observations.

A causative role for HPV infection in the wild-type p53 protein overexpression seen in inverted papillomas is well recognised: complexing of the HPV oncoprotein E6 to wild-type p53 protein is followed by stabilisation and overexpression of the p53 protein in benign inverted papillomas infected with HPV types 6/11.^34,35 In contrast, degradation of these complexes followed by negative immunostaining for p53 protein is present in malignant sinonasal lesions infected with HPV types 16/18.^35,36 In agreement with previous reports, we found HPV infection in inverted papillomas but not non-papillomatous nasal mucosal specimens.^35,36 However, the individual HPV types were not subclassified in our study: the antibody NCL-PVp detects all human HPV types.¹⁸ Molecular techniques, such as the polymerase chain reaction, are required for reliable HPV subclassification.^35,37 However, these techniques require fresh tissue,^35,37 which was not available from the cases used in our study. Hence, further investigation will be necessary to clarify the relation between distinct HPV types and p21^waf1/cip1 expression in inverted papillomas.

More p21^waf1/cip1 positive nuclei were found in transitional and squamous epithelium than in columnar epithelium in inverted papillomas. A consecutive series of histological changes has been demonstrated in inverted papillomas: columnar epithelium is gradually transformed to transitional epithelium and finally squamous epithelium.^14,38 Our findings indicate increasing p21^waf1/cip1 expression along this metaplastic process. We have recently identified increasing expression rates of p53 protein along this pathway, when investigating another cohort of inverted papillomas (manuscript in press). It is thought that p21^waf1/cip1 plays a fundamental role in the suppression of papillomatous and metaplastic progression in epithelial tissues (such as the epidermis).^39,40 Hence, p21^waf1/cip1 may be involved in the regulation of metaplasia in inverted papillomas. Current studies suggest that p21^waf1/cip1 has an inhibitory effect on viral DNA replication in HPV infected tissues, such as oral or genital papillomas.⁴¹ Further investigation is required to specify the detailed function of p21^waf1/cip1 in inverted papillomas.

In cylindrical cell papillomas, p21^waf1/cip1 expression was found only in surface cells of the epithelium. Thus, p21^waf1/cip1 expression probably occurs along with terminal differentiation. HPV infection and p53 protein overexpression were present in all specimens. However, p21^waf1/cip1 did not colocalise with p53 protein expression in basal/parabasal cells in cylindrical cell papillomas. Hence, activation of the waf1/cip1 gene is probably p53 independent in this group of endophytic sinonasal papillomas. The different behaviour of inverted papillomas and cylindrical cell papillomas in p21^waf1/cip1 expression cannot be explained from our data. A few previous studies have investigated cell cycle control in cylindrical cell papillomas, but none of those reports identified significant differences between inverted papillomas and cylindrical cell papillomas.^37,38 Further investigations are needed to determine the basic mechanisms of cell cycle control in cylindrical cell papillomas.

In squamous cell carcinomas associated with endophytic sinonasal papillomas, p21^waf1/cip1 expression was seen in one of five specimens, although p53 protein overexpression was found in all cases. Several concepts for heterogeneity in p21^waf1/cip1 expression between individual tumours have recently been presented.^42–45 For instance, mutant p53 protein is often unable to activate the transcription of the waf1/cip1 gene in breast and gastric carcinomas with p53 gene mutations.^24,41,42 Low p21^waf1/cip1 expression is associated with poor prognosis in breast carcinomas.^41,42 Sinonasal carcinomas almost exclusively express mutant p53 protein, whereas wild-type p53 protein predominates in benign sinonasal lesions.³³ Hence, the absence of p21^waf1/cip1 expression in sinonasal carcinomas probably results from mutant p53 protein expression. However, in other neoplasms such as malignant melanomas or pancreatic carcinomas p21^waf1/cip1 expression does not correlate with p53 functional status.^44,45 Transcription of the waf1/cip1 gene independent of p53 controlled pathways and post-transcriptional regulation of p21^waf1/cip1 accumulation have been demonstrated in these lesions.^44–47 Furthermore, recent studies have reported that cyclin D1 overexpression can regulate p21^waf1/cip1 expression in several human malignancies, including head and neck squamous cell carcinomas.⁴⁸ Hence, the exact regulatory mechanisms for p21^waf1/cip1 expression in sinonasal squamous cell carcinomas cannot be determined from our data. Additional investigations involving p53 gene and protein alterations, waf1/cip1 gene transcription, post-transcriptional p21^waf1/cip1 modification, and cyclin D1 overexpression are necessary to shed light on p21^waf1/cip1 expression and its prognostic value in sinonasal carcinomas.