Cytokeratin 8/18 expression indicates a poor prognosis in squamous cell carcinomas of the oral cavity

Background Intermediary filaments are involved in cell motility and cancer progression. In a variety of organs, the expression of distinct intermediary filaments are associated with patient prognosis. In this study, we seeked to define the prognostic potential of cytokeratin and vimentin expression patterns in squamous cell carcinomas (SCC's) of the oral cavity. Methods 308 patients with histologically proven and surgically treated squamous cell carcinomas of the oral cavity were investigated for the immunohistochemical expression of a variety of intermediary filaments including high- and low-molecular weight cytokeratins (Ck's), such as Ck 5/6, Ck 8/18, Ck 1, CK 10, Ck 14, Ck 19 and vimentin, using the tissue microarray technique. Correlations between clinical features and the expression of Cytokeratins and vimentin were evaluated statistically by Kaplan-Meier curves and multivariate Cox regression analysis. Results The expression of Ck 8/18 and Ck 19 were overall significantly correlated with a poor clinical prognosis (Ck 8/18 p = 0.04; Ck19 p < 0.01). These findings could also be reproduced for Ck 8/18 in primary nodal-negative SCC's and held true in multivariate-analysis. No significant correlation with patient prognosis could be found for the expression of the other cytokeratins and for vimentin. Conclusion The expression of Ck 8/18 in SCC's of the oral cavity is an independent prognostic marker and indicates a decreased overall and progression free survival. These results provide an extended knowledge about the role of intermediary filament expression patterns in SCC's.


Background
Intermediary filaments, like cytokeratins are essential intracellular components, underlying or reflecting distinct cellular properties and differentiation stages in epithelial organs. The proteins of the cytokeratin family are epithelium specific expressed [1] as low and high-molecular weight, resp. acid and basic polypeptides. Squamous, stratified epithelium is usually characterized by the expression of CK 5, which is found mainly in the basal cell layers and Ck 5 is associated with the proliferative potential of these cells [2,3]. The intermediary cell layers show an additional expression of Ck's 1 and 10, which are regarded as signs of cellular differentiation [2]. In contrast, glandular epithelia reveal the expression of lowmolecular weight cytokeratins Ck's 8/18 and 19 as typical features -expression pattern of Ck 8/18 is rather uncommon in mature squamous epithelium. Ck 19 expressed heterogeneously in the basal cell layers of stratified squamous epithelium [2,3]. Suprabasal expression of Ck 19 seems to be correlated with premalignant transformation in oral epithelium [4]. The expression of vimentin is mainly regarded as sign of a mesenchymal differentiation. However, vimentin-positively has repeatedly reported in various carcinomas and was interpreted as sign of an epithelial-mesenchymal transition, indicating an increased metastatic potential [5][6][7][8][9].
Recently, it could be shown that the induction of Ck 8/18 expression in non-malignant buccal mucosa cells resulted in a significant change of phenotypic characteristics after Ck 8/18 transfection [10]. These changes included an increased cellular motility, which might give first hints for an increased tumour aggressiveness and poor patient prognosis. However, these findings have not been transferred on a clinical level, yet. Therefore, the significance of these findings in a clinical context, especially in regard to the prognostic importance is unclear.
It was therefore the aim of this study to evaluate the expression of intermediary filaments in SCC's in relation to clinico-pathological features, especially for prognostic purposes by the use of tissue micro arrays and immunohistochemistry.

Methods
Patients with histologically proven squamous cell carcinoma of the oral floor treated surgically were eligible for the study. All patients underwent surgical treatment including radical resection of the whole tumour with a free histopathological margin of at least 4 mm from the tumour borders. Selective neck dissection of Level I, II, III and V was performed in case of suspect results in preoperative tumourstaging by computertomography or sonography examination or in case of tumour size over 2 cm, bilateral selective neck dissection was performed when the tumour spread the midline (according to the recommendation of Robins et al. 2002 [11]). Radiotherapy was performed when lymph node metastases were detected histologically. All tumours were classified postsurgically according to the TNM system 6 th eds. UICC 2002 [12].
For tissue micro array (TMA) construction, two punch biopsies with a diameter of 0.6 mm from each donor block were taken and transferred into the new acceptor block. TMA construction was performed by using a special tissue micro array instrument (Beecher Instruments, New Jersey, USA), according to standard protocols [13,14].
Immunohistochemistry was performed on 4-µm-thick TMA sections. After deparaffinization and rehydration, endogenous peroxidase activity was blocked for 30 minutes in methanol containing 0.3% hydrogen peroxide. After antigen retrieval, a cooling-off period of 20 minutes preceded the incubation of the primary antibody. Thereafter, the catalyzed signal amplification system (DAKO, Glostrup, Denmark) was used for Ck 5/6, Ck 8/18, Ck 1, CK 10, Ck 14, Ck 19 and vimentin staining according to the manufacturer's instructions.
All antibodies were detected by a standard avidin-biotin complex method with a biotinylated rabbit anti-mouse antibody (DAKO) and an avidin-biotin complex (DAKO). The stainings were developed with diaminobenzidine or alternatively using LSAB/AP (DAKO). Before the slides were mounted, all sections were counterstained for 45 seconds with hematoxylin and dehydrated in alcohol and xylene. Appropriate negative controls (obtained by omis- sion of the primary antibody) and positive controls were used throughout.
The expression of Ck 5/6, Ck 8/18, Ck 1, CK 10, Ck 14, Ck 19 and vimentin were determined independently by two pathologists (HB and JP). Both pathologists determined the percentage of positive cells in each core.

Statistical analysis
The TNM-stage, the histological differentiation and the expression of Ck 5/6, Ck 8/18, Ck 1, CK 10, Ck 14, Ck 19 and vimentin were related to the duration of the progression-free and the overall survival. The measurement of time started from the date of surgery to the date of histologically proven recurrent or metastatic carcinoma or disease related death, respectively. Patients who died from intercurrent disease during the trial were censored at the date of death. Patients lost to follow-up were censored at the date of the last examination.
Correlations between clinical features and the expression of cytokeratins were evaluated statistically by chi-square tests. The overall survival curves were constructed according to Kaplan and Meier [15]. The Log-Rank test was used to assess differences between groups and the multivariate survival analysis was performed with Cox regression [16], a p-value < 0.05 was considered to be significant.

Results
The series comprised 308 patients (240 men and 68 women) with a median age of 58 years (range 31-90). All tumours were classified post surgically according to the TNM system [12]. Patients were clinically evaluated in our routine follow-up for 3-10 years. Clinical and tumour details of the patients with oral squamous cell cancer are shown in Table 2.
On average 90 % of all cores could be evaluated for the expression of intermediate filaments.  Table 3). The immunohistochemical staining is shown in Figure 1.
Ck 5/6 and Ck 14 were significantly correlated with the tumour size and lymph node status in the Chi-square Test (    Immunohistochemical staining of oral squamous cell carcinoma with cytokeratin cytokeratin antibodies All carcinomas with an expression of Ck 8/18 and Ck 19 were reviewed and re-evaluated on whole sections. This was done in order to exclude the possibility of a misinterpretation and misclassification of dedifferentiated salivary gland carcinomas as poorly differentiated SCC's. However, in all these tumours the diagnosis of a primary salivary gland carcinoma could be ruled out. In multivariate Cox analysis we compared the overall survival according to clinically established prognostic factors tumour size and nodal status with the expression of Ck 8/ 18, Ck 5/6, Ck 1, CK 10, Ck 14 and Ck 19 (Table 5). Three parameters as nodal status, tumour size and Ck 8/18 expression were indicated by Cox regression as independent predictors of survival prognosis.

Discussion
Intermediate filaments are central components of the intracellular skeleton. The expression of distinct intermediate filament proteins is tissue-specific and almost highly conserved during carcinogenesis. Intermediate filaments are interpreted as mere bystanders and markers of distinct intracellular, cell-specific regulation mechanisms. Several studies associated changes in intermediate filament expression with an altered cellular behaviour [17,18].
The expression of high molecular weight cytokeratins, especially Ck 5 is a hallmark of squamous epithelium and is predominantly seen in the basal layers of stratified epithelium [2,19]. This cell layer is regarded as the anatomical localization of tissue specific stem/progenitor cells. Stem cells of stratified epithelium have been repeatedly described as the major cellular targets for cancer causing mutations and therefore might give in a long term rise to the development of SCC's. In the context of the existing literature it seems logical that SCC's retain and are pheno- Our study shows in a series of more than 300, uniformly treated, primary SCC's of the oral cavity that the expres-sion of glandular cytokeratins such as Ck 8, 18 and 19, pinpoints a significant subgroup of SCC's. These carcinomas were associated with a poor prognosis, independently of the primary, preoperative lymph node status. It is important to stress that this held also true in multivariate analysis. This might be surprising on a first glance, since recent other clinico-pathological studies in SCC's of the oral cavity, especially with the focus on Ck 19 revealed inhomogeneous results [21][22][23][24]. This might on the one hand be due to significantly lower numbers of investigated tumour samples, and on the other hand caused by different techniques for the determination for the expres-   In general, cytokeratin expression patterns are highly-conserved. The above mentioned data from the literature showed that any change in the intermediate filament expression pattern reflects or underlies changes in substantial cellular properties. Strong homologies could also be shown in other carcinomas. In primary breast cancer, this could be demonstrated with different approaches. Down-regulated expression of cytokeratin Ck 8/18 was associated with increased metastatic properties and poor prognosis in breast cancer [30]. The expression of Ck 5 in a small subset of invasive breast cancers was in contrast associated with a poor prognosis [31].
Interestingly, the expression of vimentin, alone or in combination with Ck 8/18 and 19, was not associated with poor prognosis or higher tumour stages. A multitude of studies gave evidence that the expression of vimentin in carcinomas has to be interpreted as a sign of an epithelialmesenchymal transition, associated with high tumour aggressiveness [9,32]. This association remains questionable, since the expression of vimentin has already been described in non-invasive premalignant lesions of squamous epithelium and cultured keratinocytes.

Conclusion
In conclusion, our results in a series of 308 primary, previously untreated SCC's of the oral cavity showed an important prognostic impact for the expression of glandular, simple cytokeratins in SCC's. Further studies have to unravel the underlying regulatory changes in order to allow the definition of specific therapies of these high risk cancer patients.