The role of miRNAs in human papilloma virus (HPV)-associated cancers: bridging between HPV-related head and neck cancer and cervical cancer.

This corrects the article DOI: 10.1038/bjc.2012.109.

Increasing incidence of base of tongue cancers from 2000 to 2010 due to HPV: the largest demographic study of 210 Danish patients.

BACKGROUND: We assessed the development in the number of new base of tongue squamous-cell carcinoma (BSCC) cases per year in eastern Denmark from 2000 to 2010 and whether HPV may explain any observable increased incidence. METHODS: We performed HPV DNA PCR and p16 immunohistochemistry analysis for all (n=210) BSCCs registered in the Danish Head and Neck Cancer Group (DAHANCA) and the Danish Pathology Data Bank, and genotyped all HPV-positive specimens with amplicon-based next-generation sequencing. RESULTS: The overall crude incidence of BSCCs increased significantly (5.4% per year) during the study period. This was explained by a significant increase in the number of HPV-positive BSCCs (8.1% per year), whereas the number of HPV-negative BSCCs did not increase significantly. The overall HPV prevalence was 51%, with HPV16 as the predominant HPV type. CONCLUSIONS: The increased number of HPV-positive BSCCs may explain the increasing incidence of BSCCs in eastern Denmark, 2000-2010.

Molecular profiling of tumour budding implicates TGFß-mediated epithelial-mesenchymal transition as a therapeutic target in oral squamous cell carcinoma.

Although tumour budding is an adverse prognostic factor for many cancer types, the molecular mechanisms governing this phenomenon are incompletely understood. Therefore, understanding the molecular basis of tumour budding may provide new therapeutic and diagnostic options. We employ digital image analysis to demonstrate that the number of tumour buds in cytokeratin-stained sections correlates with patients having lymph node metastases at diagnosis. The tumour bud count was also a predictor of overall survival, independent of TNM stage. Tumour buds and paired central tumour areas were subsequently collected from oral squamous cell carcinoma (OSCC) specimens, using laser capture microdissection, and examined with RNA sequencing and miRNA-qPCR arrays. Compared with cells from the central parts of the tumours, budding cells exhibited a particular gene expression signature, comprising factors involved in epithelial-mesenchymal transition (EMT) and activated TGFß signalling. Transcription factors ZEB1 and PRRX1 were up-regulated concomitantly with the decreased expression of mesenchymal-epithelial (MET) transcription factors (eg OVOL1) in addition to Krüppel-like factors and Grainyhead-like factors. Moreover, miR-200 family members were down-regulated in budding tumour cells. We used immunohistochemistry to validate five markers of the EMT/MET process in 199 OSCC tumours, as well as in situ hybridization in 20 OSCC samples. Given the strong relationship between tumour budding and the development of lymph node metastases and an adverse prognosis, therapeutics based on inhibiting the activation of TGFß signalling may prove useful in the treatment of OSCC.

Correlation between human papillomavirus and p16 overexpression in oropharyngeal tumours: a systematic review.

BACKGROUND: A significant proportion of squamous cell carcinomas of the oropharynx (OP-SCC) are related to human papillomavirus (HPV) infection and p16 overexpression. This subgroup proves better prognosis and survival but no evidence exists on the correlation between HPV and p16 overexpression based on diagnostic measures and definition of p16 overexpression. We evaluated means of p16 and HPV diagnostics, and quantified overexpression of p16 in HPV-positive and -negative OP-SCCs by mode of immunohistochemical staining of carcinoma cells. METHODS: PubMed, Embase, and the Cochrane Library were searched from 1980 until October 2012. We applied the following inclusion criteria: a minimum of 20 cases of site-specific OP-SCCs, and HPV and p16 results present. Studies were categorised into three groups based on their definition of p16 overexpression: verbal definition, nuclear and cytoplasmatic staining between 5 and 69%, and ≥70% staining. RESULTS: We identified 39 studies with available outcome data (n=3926): 22 studies (n=1980) used PCR, 6 studies (n=688) used ISH, and 11 studies (n=1258) used both PCR and ISH for HPV diagnostics. The methods showed similar HPV-positive results. Overall, 52.5% of the cases (n=2062) were HPV positive. As to p16 overexpression, 17 studies (n=1684) used a minimum of 5-69% staining, and 7 studies (n=764) used ≥70% staining. Fifteen studies (n=1478) referred to a verbal definition. Studies showed high heterogeneity in diagnostics of HPV and definition of p16. The correlation between HPV positivity and p16 overexpression proved best numerically in the group applying ≥70% staining for p16 overexpression. The group with verbal definitions had a significantly lower false-positive rate, but along with the group applying 5-69% staining showed a worse sensitivity compared with ≥70% staining. CONCLUSIONS: There are substantial differences in how studies diagnose HPV and define p16 overexpression. Numerically, p16 staining is better to predict the presence of HPV (i.e. larger sensitivity), when the cutoff is set at ≥70% of cytoplasmatic and nuclear staining.

Ten-year retrospective study of head and neck carcinoma in situ: incidence, treatment, and clinical outcome.

OBJECTIVES: To examine the management and clinical outcome for patients with primary head and neck carcinoma in situ (CIS) and to estimate the incidence in the referral population. STUDY DESIGN: A retrospective study from 2000-2009 of patients with head and neck CIS referred for treatment at Rigshospitalet. The referral area was East Denmark and Greenland with a population of 2.4 million. RESULTS: Fifty-five patients with primary CIS were identified: 21 oral cavity, 7 pharynx, 25 larynx, 2 nasal cavity/paranasal sinuses. The median annual incidence was 0.24/100,000. Eleven patients (20%) had T-site recurrence. The 5-year disease-specific survival rate and 5-year recurrence-free survival rate were 98% and 74% respectively. CONCLUSIONS: The annual incidence of primary head and neck CIS was low and in accordance with previous findings reported in the literature. We recommend that CIS lesions should be treated on T-site and surveilled as T1/T2 head and neck carcinomas.

The role of miRNAs in human papilloma virus (HPV)-associated cancers: bridging between HPV-related head and neck cancer and cervical cancer.

BACKGROUND: Although the role of human papilloma virus (HPV) in cervical squamous cell carcinoma (CSCC) is well established, the role in head and neck SCC (HNSCC) is less clear. MicroRNAs (miRNAs) have a role in the cancer development, and HPV status may affect the miRNA expression pattern in HNSCC. To explore the influence of HPV in HNSCC, we made a comparative miRNA profile of HPV-positive (HPV+) and HPV-negative (HPV-) HNSCC against CSCC. METHODS: Fresh frozen and laser microdissected-paraffin-embedded samples obtained from patients with HPV+/HPV- HNSCC, CSCC and controls were used for microarray analysis. Differentially expressed miRNAs in the HPV+ and HPV- HNSCC samples were compared with the differentially expressed miRNAs in the CSCC samples. RESULTS: Human papilloma virus positive (+) HNSCC had a distinct miRNA profile compared with HPV- HNSCC. Significantly more similarity was seen between HPV+ HNSCC and CSCC than HPV- and CSCC. A set of HPV core miRNAs were identified. Of these especially the miR-15a/miR-16/miR195/miR-497 family, miR-143/miR-145 and the miR-106-363 cluster appear to be important within the known HPV pathogenesis. CONCLUSION: This study adds new knowledge to the known pathogenic pathways of HPV and substantiates the oncogenic role of HPV in subsets of HNSCCs.

Different miRNA signatures of oral and pharyngeal squamous cell carcinomas: a prospective translational study.

BACKGROUND: MicroRNAs (miRNAs) are small non-coding RNAs, which regulate mRNA translation/decay, and may serve as biomarkers. We characterised the expression of miRNAs in clinically sampled oral and pharyngeal squamous cell carcinoma (OSCC and PSCC) and described the influence of human papilloma virus (HPV). METHODS: Biopsies obtained from 51 patients with OSCC/PSCC and 40 control patients were used for microarray analysis. The results were correlated to clinical data and HPV status. Supervised learning by support vector machines was employed to generate a diagnostic miRNA signature. RESULTS: One hundred and fourteen miRNAs were differentially expressed between OSCC and normal oral epithelium, with the downregulation of miR-375 and upregulation of miR-31 as the most significant aberrations. Pharyngeal squamous cell carcinoma exhibited 38 differentially expressed miRNAs compared with normal pharyngeal epithelium. Differences in the miRNA expression pattern of both normal epithelium and SCC were observed between the oral cavity compared with the pharynx. Human papilloma virus infection revealed perturbations of 21 miRNAs, most significantly in miR-127-3p and miR363. A molecular classifier including 61 miRNAs was generated for OSCC with an accuracy of 93%. CONCLUSION: MicroRNAs may serve as useful biomarkers in OSCC and PSCC. The influence of HPV on miRNA may provide a mechanism for the distinct clinical behaviour of HPV-infected tumours.

Expression of two drug-metabolizing cytochrome P450-enzymes in human salivary glands.

OBJECTIVE: The oral cavity is constantly lubricated by saliva and even small amounts of xenobiotics and / or their metabolites in the saliva may affect the oral mucosa. Our aim was therefore to clarify if xenobiotic metabolizing enzymes CYP1A2 and CYP3A4 are expressed in salivary glands. METHODS: Formalin-fixed paraffin-embedded specimens from parotid (10), submandibular (7) and labial (10) salivary glands were examined immunohistochemically and by in situ hybridization for expression of CYP1A2 and CYP3A4 protein and mRNA. RESULTS: CYP1A2 and CYP3A4 protein and mRNA were detected in ductal and seromucous / serous acinar cells in all gland types although to a varying degree and intensity. Mucous acinar cells were positive to a lesser extent. CONCLUSION: The results indicate a xenobiotic metabolizing capability of salivary glands. This may have implications for development of oral mucosal disease as a result of mucosal exposure to metabolites originating from internal sources (blood) as well as from saliva.

Expression of polypeptide GalNAc-transferases in stratified epithelia and squamous cell carcinomas: immunohistological evaluation using monoclonal antibodies to three members of the GalNAc-transferase family.

Mucin-type O-glycosylation is initiated by a large family of UDP-GalNAc: polypeptide N -acetyl-galactosaminyltransferases (GalNAc-transferases). Individual GalNAc-transferases appear to have different functions and Northern analysis indicates that they are differently expressed in different organs. This suggests that O-glycosylation may vary with the repertoire of GalNAc-transferases expressed in a given cell. In order to study the repertoire of GalNAc-transferases in situ in tissues and changes in tumors, we have generated a panel of monoclonal antibodies (MAbs) with well defined specificity for human GalNAc-T1, -T2, and -T3. Application of this panel of novel antibodies revealed that GalNAc- transferases are differentially expressed in different cell lines, in spermatozoa, and in oral mucosa and carcinomas. For example, GalNAc-T1 and -T2 but not -T3 were highly expressed in WI38 cells, and GalNAc-T3 but not GalNAc-T1 or -T2 was expressed in spermatozoa. The expression patterns in normal oral mucosa were found to vary with cell differentiation, and for GalNAc-T2 and -T3 this was reflected in oral squamous cell carcinomas. The expression pattern of GalNAc-T1 was on the other hand changed in tumors to either total loss or expression in cytological poorly differentiated tumor cells, where the normal undifferentiated cells lacked expression. These results demonstrate that the repertoire of GalNAc-transferases is different in different cell types and vary with cellular differentiation, and malignant transformation. The implication of this is not yet fully understood, but it suggests that specific changes in sites of O-glycosylation of proteins may occur as a result of changes in the repertoire of GalNAc-transferases.

Salivary gland carcinomas: prognostic significance of simple mucin-type carbohydrate antigens.

The prognosis of salivary gland carcinomas is difficult to assess. Simple mucin-type carbohydrates (T and sialosyl-T antigens, Tn and sialosyl-Tn antigens) have been shown to be of value in predicting prognosis for carcinomas in other locations. We studied the prognostic significance of the expression of these structures in a retrospective study of 133 patients with salivary gland carcinomas, using immunohistochemistry and a panel of well-defined monoclonal antibodies (MAbs) on formalin-fixed paraffin-embedded tissues. Sialosyl-Tn, T and sialosyl-T antigens were not correlated with prognosis. Univariate analyses showed no overall difference in survival or locoregional control between patients with Tn-positive and patients with Tn-negative tumours, but indicated that expression of the Tn antigen was associated with early locoregional recurrences and deaths. Tn was, however, not an independent prognostic factor by multivariate regression analysis.

Can alterations in integrin and laminin-5 expression be used as markers of malignancy?

Development of squamous cell carcinomas (SCCs) involves alterations in the adhesive interactions in the epithelium and invasion through the basement membrane. Therefore, changes in the expression of receptors and ligands involved in cell-cell and cell-matrix adhesion may be essential for the transformation of a premalignant into a malignant lesion. The aim of this study was to examine if expression of specific cell adhesion molecules can be used as markers of malignant development. By immunohistochemistry, we examined the expression pattern of integrins alpha2beta1, alpha3beta1, alpha6beta4 and laminin-5 in biopsies from SCCs (n=18), premalignant lesions (leukoplakias, n=21) and non-premalignant tissue with chronic inflammation (n=11). In poorly differentiated SCCs, patchy loss of alpha3beta1, alpha6beta4 and laminin-5 expression was pronounced at the invasion front, whereas there was a tendency to increased expression of alpha2beta1. Analogous to the SCCs, biopsies from the leukoplakias and the non-premalignant inflammatory tissue showed alterations of the expression of alpha3beta1 and alpha6beta4 in the basal cell layers and of laminin-5. However, a characteristic finding in biopsies from leukoplakias was loss of alpha2beta1 and alpha3beta1 in the suprabasal cells. There was no unequivocal expression of the adhesion molecules distinguishing between inflammatory tissue, premalignant, and malignant lesions.

Salivary gland carcinomas--prognostic factors.

A retrospective study of factors of prognostic significance for clinical course and survival was performed using uni- and multivariate analyses in 251 patients with primary salivary gland carcinoma admitted during the period 1958-1992. Univariate analyses indicated that site of primary tumour, histology, clinical stage, presence of node metastases at primary diagnosis, and status of surgical margins were important prognostic factors for cause-specific survival, locoregional control and distant metastases. Multivariate analyses confirmed that histology was important for both locoregional control and cause-specific survival, whereas primary site was only of importance for locoregional control. Presence of node metastases at diagnosis was more important for locoregional control than clinical stage, whereas clinical stage was the most important factor for cause-specific survival. Status of surgical margins was of major importance for both cause-specific survival and locoregional control. Radiotherapy in addition to surgery improved locoregional control only, whereas survival was not affected.

Observer variability in histological malignancy grading of adenoid cystic carcinomas.

The value of malignancy grading of adenoid cystic carcinomas (ACC) is controversial. Some studies have shown that tumours with a solid growth component have a rapid fatal course, compared to tumours without a solid growth component, in which recurrences develop even many years after initial treatment. Other studies have failed to correlate growth patterns with clinical course. No universally accepted grading system exists and no reproducibility studies of the existing grading systems have been performed. The aim of this study was to examine the reproducibility of grading based on semi-quantitative assessment of the solid growth pattern in ACC. Two different grading systems were assessed by 3 observers on a material of 59 ACC. Interobserver agreement was evaluated using the kappa statistic. The reproducibility of grading was poor, except for the category "solid component constituting 50% or more of the tumour" (kappa = 0.52). It is concluded that quantitative methods are necessary if grading is to be used in prognostic evaluation of ACC. The rarity of the tumours, however, combined with difficulties in diagnosis will impede such investigations unless multicentre studies are undertaken.

Identification of a major human high molecular weight salivary mucin (MG1) as tracheobronchial mucin MUC5B.

Human saliva contains high and low molecular weight mucin glycoproteins, that are distinct. Recently the gene encoding low molecular weight salivary mucin was cloned and designated MUC7, whereas the primary structure of high molecular weight salivary mucin is unclear. Furthermore, the expression patterns of high and low molecular weight salivary mucins in salivary glands have been debated. We have previously generated monoclonal antibodies specific for the peptide cores of salivary mucins. In the present study a monoclonal antibody specific for high molecular weight salivary mucin was used to screen a human salivary gland cDNA library. A single clone, SAL1, was identified and found to be encoded by tracheobronchial mucin gene MUC5B. A previously reported partial cDNA sequence from salivary mucin was linked to SAL1/MUC5B by genomic cloning and reverse transcriptase-polymerase chain reaction. Northern analysis of salivary gland RNA probed with SAL1 suggested that MUC5B was highly expressed in salivary glands. In situ hybridization was performed with a SAL1/MUC5B probe and a MUC7 probe. All mucous cells from the submandibular, sublingual, palatine, and labial glands labeled with the MUC5B probe, while serous cells labeled with the MUC7 probe. These findings were in accordance with our previous immunohistological results of the cellular localizations of salivary mucins. The results suggest that MUC5B is identical to or a major fraction of high molecular weight salivary mucin, and that MUC5B is expressed in all mucous cells of salivary glands. In contrast MUC7 is expressed in serous cells of salivary glands except the parotid glands.

Loss of a novel mucin-like epithelial glycoprotein in oral and cervical squamous cell carcinomas.

Stratified squamous epithelia of oral and cervical mucosa express high levels of simple mucin-type O-linked carbohydrates, and these are known to undergo structural changes in relation to epithelial differentiation and neoplastic transformation. O-glycans in these epithelia are associated with the cell membrane, but the identity of the carrier molecule(s) remains largely unknown. We report here the identification of a membrane-bound M(r) 200,000-250,000 glycoprotein (gp230) that is expressed in stratified squamous epithelia of the oral cavity. Western blot analysis identified gp230 as a major carrier of simple-mucin type carbohydrate antigens in buccal nonkeratinized mucosal epithelium, suggesting that it may represent a mucin-like molecule. A monoclonal antibody PANH4 defining a protein epitope of gp230 was generated. The PANH4 epitope was localized by immunohistology to suprabasal cell layers of buccal epithelium and was also found in larynx, esophagus, vagina, and exocervix, but not in epidermis. Data showed that gp230 was distinct from MUC1 or CD44. It is interesting that in most cases gp230 was not expressed in squamous cell carcinomas of buccal and cervical mucosa. A few moderately differentiated carcinomas, mainly from cervix, expressed the gp230 epitope. The results suggest that a membrane-bound mucin-like molecule, gp230, is associated with the differentiated phenotype of normal mucosal stratified squamous epithelia and that expression of gp230 generally is lost in severe oral epithelial dysplasia and squamous cell carcinomas of oral and cervical mucosa.

Differential expression of human high-molecular-weight salivary mucin (MG1) and low-molecular-weight salivary mucin (MG2).

Two distinct mucin components of saliva, MG1 and MG2, have been identified based on chemical composition and molecular weights (high and low, respectively) in saliva. With the aim of characterizing the expression pattern of salivary mucins, we have prepared monoclonal antibodies (MAbs) directed against the peptide core of MG1 and against a synthetic peptide derived from the MG2 (MUC7) sequence. MAb PANH2 raised against partially deglycosylated MG1 stained a high-molecular-weight smear in Western blots of partially purified MG1. PANH2 binding was increased by deglycosylation with trifluoromethanesulfonic acid as well as with subsequent periodate treatment, and was eliminated by pronase treatment, strongly suggesting that MAb PANH2 was directed to a peptide epitope of MG1. MAb PANH3 raised against a synthetic peptide derived from the MG2 (MUC7) sequence reacted with the native molecule and stained a narrow smear of ca. 200,000 to 210,000 in Western blots of concentrated saliva and a lower-molecular-weight smear of trifluoromethanesulfonic-acid-treated MG2. Immunohistology on frozen sections of human salivary glands showed that MAb PANH2 selectively labeled mucous cells, whereas MAb PANH3 labeled subpopulations of serous cells. Double-direct immunofluorescence staining with PANH2 and PANH3 demonstrated that the staining patterns were non-overlapping. The development of these antibody probes will facilitate studies of mucin expression in diseases of salivary glands.

A sensitive enzyme-linked immunosorbent assay used for quantitation of epidermal growth factor receptor protein in head and neck carcinomas: evaluation, interpretations and limitations.

The EGF receptor is a transmembrane glycoprotein exerting mitogenic effects on epithelial cells. The purpose of the present study was to develop a sensitive enzyme-linked immunosorbent assay (ELISA) for determination of the epidermal growth factor receptor (EGFR) protein to examine whether the receptor was overexpressed in head and neck squamous cell carcinomas compared with the normal counterpart, and to establish whether clinicopathological correlations were present by investigating a broad spectrum of parameters (tumour size, clinical stage, positive lymph nodes, tumour site, histological grade, keratinisation, preoperative irradiation and clinical outcome). The assay employs two commercially available monoclonal antibodies, both detecting protein epitopes. The material comprises 60 head and neck carcinomas, corresponding normal tissue and normal oral mucosa from healthy individuals. The study demonstrates significantly higher receptor levels in tumours compared with normal tissue (P < 0.002) and a range in tumours and normal tissues of 0.4-10.5 and 0.1-4.3 nmol g-1 membrane protein respectively. Quantitation of receptors in normal mucosa emphasises the importance of using the patients' corresponding normal tissue, because using the patients' mucosa resulted in 83% overexpression, while using normal mucosa from healthy individuals only demonstrated overexpression in 50% of cases. No significant clinicopathological correlations could be established, although the mean values for EGFR increased with tumour size and advanced clinical stage. Furthermore, the prognostic value concerning disease-free survival, recurrence and the time interval for recurrence were investigated but no significance could be demonstrated. In conclusion, the investigation supports the theory of overexpression of EGFR protein as a common motif for malignant epithelial tumours, but limitations in interpretations are demonstrated and discussed further.

The Thomsen-Friedenreich (T) simple mucin-type carbohydrate antigen in salivary gland carcinomas.

The simple mucin-type T (Thomsen-Friedenreich) antigen is a marker of carcinomas, and has been related to aggressiveness of malignant tumours. We studied the expression of T, sialosyl-T, A and H blood group antigens in salivary gland carcinomas. The aim was to study whether the tumours, based on the expression of these structures, could be divided into new diagnostic groups that may later show prognostic significance. Formalin fixed paraffin-embedded tissue sections from 77 salivary gland carcinomas of different histological types were studied using immunohistology and monoclonal antibodies (MAbs). Fresh frozen tissue was examined in 30 of the cases. Frozen sections were superior to paraffin sections in demonstrating T and H antigens. Aberrant glycosylation with accumulation of T (in cytoplasm) and sialosyl-T antigens (in cytoplasm, membrane and mucin) was found in all tumour types except acinic cell carcinomas. In carcinomas in pleomorphic adenomas (CinPA) the effect of fixation was minimal and T antigen location was different. In carcinomas with myoepithelial cell (MEC) participation, the MECs had retained a normal glycosylation pattern. H antigen was expressed in all tumour types, except acinic cell carcinomas and CinPA. A antigen was expressed in all tumour types from blood group A patients, except in CinPA. The expression of T, sialosyl-T, H and A antigens in relation to differentiation grade varied with tumour type in poorly differentiated areas. High and moderate differentiated areas were always stained, whereas poorly differentiated areas in some tumour types expressed T and sialosyl-T antigens and others did not. The accumulation versus lack of expression of the investigated structures in poorly differentiated areas of the tumours may be of prognostic significance.

Thomsen-Friedenreich (T) antigen as marker of myoepithelial and basal cells in the parotid gland, pleomorphic adenomas and adenoid cystic carcinomas. An immunohistological comparison between T and sialosyl-T antigens, alpha-smooth muscle actin and cytokeratin 14.

Controversy centres on the role and identification of myoepithelial (MEC) and basal cells in salivary gland tumours, and recent studies suggest that both basal cells and myoepithelial cells participate in the formation of salivary gland tumours. We have correlated the expression of different well-known markers of normal MEC/basal cells (i.e. alpha-smooth muscle actin and cytokeratin 14) with T (Thomsen-Friedenreich) antigen and its sialylated derivative: sialosyl-T antigen,) in 17 normal parotid glands and in two tumour types with MEC participation (i.e pleomorphic adenomas (PA) and adenoid cystic carcinomas (ACC)) using immunohistology with well-defined monoclonal antibodies (MAbs). Paraffin-embedded/fresh frozen tissue sections were studied from 33/17 patients with PA and 15/7 patients with ACC. In normal parotid tissue coexpression of alpha-smooth muscle actin, cytokeratin 14, T and sialosyl-T antigens was found in all MEC and in some of the basal cells lining striated ducts. The remaining basal cells exclusively expressed cytokeratin 14, T and sialosyl-T antigens. In the tumours, cells believed to be modified myoepithelial cells showed two different staining patterns: 1) Coexpression of alpha-smooth muscle actin, cytokeratin 14, T and sialosyl-T antigens, and 2) Coexpression of cytokeratin 14, T and sialosyl-T antigens, but no alpha-smooth muscle actin. The epithelial ductular structures in the tumours showed aberrant expression of cytokeratin 14, T and sialosyl-T antigens, and cytokeratin 14 was the only marker of cells in solid undifferentiated areas of adenoid cystic carcinomas. Our study supports the view, that modified "myoepithelial" cells in the tumours consist of a mixture of basal cells and myoepithelial cells. None of the investigated structures was in itself an ideal marker in the identification of MEC/basal cells. The cells can be identified by a combination of markers (i.e. cytokeratin 14, alpha-smooth-muscle actin, T and sialosyl-T antigens).