Human external auditory canal cholesteatoma (EACC) is not often seen in otolaryngology. Some authors have noted circulatory disorders of the local blood vessels as the etiologic factor for establishing EACC. Diminished oxygen supply results in the attempt to establish angiogenesis. Hepatocyte growth factor/scatter factor (HGF/SF) and vascular endothelial growth factor (VEGF) are the most important angiogenic factors in this process. In a recent study we described strong expression of VEGF and HGF in EACC. All EACC and normal AMS cell cultures were obtained from 5 patients undergoing surgery and used at passage 3. After 16 to 72 h of incubation with 20 ng/ml HGF/SF, the expression of the VEGF protein in the supernants of the HGF/SF-treated and untreated culture was analyzed. EACC-culture cells showed a stronger baseline expression of VEGF. After 72 h of incubation with 20 ng/ml HGF/SF of HGF/SF, the expression of VEGF in normal keratinocytes was 173.4 pg/ml. The expression level of VEGF in the EACC culture was 275.73 pg/ml. We observed a 2.5-fold induction of VEGF in EACC after 72 h, which started with 1.5-fold baseline VEGF concentrations of normal keratinocytes. Our analysis showed that, in the EACC culture, VEGF was elevated after treatment with HGF/SF. HGF/SF appears to activate cellular pathways inducing release of VEGF. After purification, no fibroblasts were present in our EACC culture so as to exclude possible paracrine effects by fibroblasts.

To date, the underlying genomic changes in benign and malignant tumors of salivary-gland and paranasal-sinus origin are poorly understood. This is due in part to the low incidence of these tumors and the enormous histological variety of tumors within this head and neck region. We examined 58 of these tumors (14 adenoid cystic carcinomas, 9 adenocarcinomas, 5 cylindrical carcinomas, 11 pleomorphic adenomas, and 19 inverted papillomas) by dual fluorescence in situ hybridization (FISH) with centromere-specific probes on six chromosomes (3, 7, 9, 11, 17, and 18) for numerical changes. In adenoid cystic carcinomas, monosomy of chromosome 17 and polysomy of chromosomes 3, 9 and 11 were most frequently encountered. In adenocarcinomas, monosomy of chromosome 17 and polysomy of chromosomes 7 and 11 were most frequent. In cylindrical cell carcinomas, polysomy of chromosomes 7, 9, 11 and 17 was present in the majority of tumors. Disomy is rare, even in benign tumors. Polysomy is more frequent in malignant tumors than in benign. Tetrasomy is found almost only in malignant tumors. In summary, the occurrence of polysomy might reflect a step towards malignancy in tumors of the salivary glands and paranasal mucosa. Polysomy of chromosome 11 could be defined as typical for all investigated histological types of malignant tumor in this region of the head and neck.

Metalloproteinases have been characterised as "destroying bulldozers" in the extracellular matrix permitting normal remodelling and contributing to pathological tissue destruction. The opposing system is established by the Cadherin-beta-catenin system, which assures the integrity of the tissue. The EACC is an extremely rare disease in the field of Otolaryngology and its incidence is estimated about 1 per 1000 new otologic patients. The aim of this study is to characterise the balance between metalloproteinases and beta-catenin in EACC tissue. Twelve specimens were obtained during surgical removal of the EACC. The EACC- and AMS-specimens were immunostained with antibodies for beta-catenin, MMP-2 and MMP-9, respectively. Immunostaining for gelatinases was increased in all layers of the EACC. However, the normal auditory meatal skin presented moderate immunostaining. In the EACC specimens, the basal layers of the matrix were positive for beta-catenin. The suprabasal layers showed diminished or negative immunostaining for beta-catenin. In all layers the AMS was homogeneously positive for beta-catenin. Metalloproteinases can modulate the balance between cellular growth and apoptosis through cleavage of non-matrix, cell-surface substrates, such as the E-Cadherin-beta-catenin system. Furthermore, this balance guarantees the integrity of the tissue. Unbalanced conditions such as described in EACC, result in unregulated desquamation and accumulation of dead keratinocytes, invasive and defective growth into adjacent tissue, and loss of growth control. Generating synthetic inhibitors of metalloproteinases for therapeutic use in EACC and other defective diseases with unbalanced tissue conditions seems useful.

Intratumoral genomic heterogeneity, which can be defined as both intersample and intrasample heterogeneity, is still a poorly understood phenomenon in head and neck squamous cell carcinoma (HNSCC) with presumed implications on tumor behavior and even prognosis. We analyzed 89 tumor specimen from 37 HNSCC patients by fluorescence in situ hybridization (dual-FISH) using specific DNA probes binding to centromeric sites of 6 chromosomes to investigate intratumoral heterogeneity. A derivation from disomy in at least 1/6 chromosomes was detected in 88/89 (99%) specimen. In 33% of these samples, a change in ploidy could be suspected. Intrasample heterogeneity was detected in 68/89 (76%). Intrasample heterogeneity was more pronounced in primary tumors than in metastatic tumors. Analysis of the intersample heterogeneity revealed notable differences between the 6 chromosomes with the highest discordance detected for chromosome 3 (46%) and the lowest for chromosome 11 (27%). Following our results, it seems important to us to underline that intratumoral heterogeneity exists as intra- and sample heterogeneity in HNSCC. Altogether, trisomic cells were significantly more frequent in primary tumors than in metastases (p=0.01) while, in turn, monosomic cells were significantly more frequent in metastases (p=0.029). In individual cases the extent of discordance between corresponding samples made a common clonal precursor unlikely. In these cases, the synchronous development of a primary tumor and a carcinoma of unknown primary ('CUP syndrome'), otherwise undetected, should be considered.

Very little is known about possible intra-tumoural genetic heterogeneity between primary tumours and lymph node metastases in head and neck squamous cell carcinoma (HNSCC). To investigate this phenomenon, we analysed 96 micro-dissected tumour samples for allelic imbalance at four of the most frequently altered chromosomal locations in HNSCC (3p14.2; 9p21; 11q23.3; 17p13.1) using microsatellite markers. From 23 patients, matched pairs of primary tumour and lymph node metastasis were analysed. Discordance in the allelic distribution was identified in 8 cases (35%). With one exception, the metastasis contained a more balanced allelic status than the primary tumour. In contrast, in a group of 25 tumours with two anatomically different samples from the primary tumour site, discordance was identified in only 3 tumours (13%). These results are compatible with the dissemination of subclones from the primary tumour site with a more balanced allelotype in the metastasis. In our opinion, several scenarios could explain this phenomenon. From a clinical point of view, genetic discordance between the metastasis and the primary tumour must be taken into consideration when establishing molecular biologic markers for choice of therapy and prognosis in head and neck cancer.

The huge majority of head and neck squamous cell carcinoma (HNSCC) show alterations of p53 either on the genetic level or on the protein level. Allelic imbalance (AI)/loss of heterozygosity (LOH) on 17p at the p53 locus is frequent in HNSCC. However, the complex relationship between these phenomena is poorly understood in HNSCC. We investigated one group of 39 HNSCC for: a) allelic imbalance on 17p using 4 microsatellite markers located throughout this chromosomal arm; b) mutations of p53 in exons 5-9; and c) overexpression of p53 using two antibodies located on opposite ends of the protein. AI/LOH was detected in 44% at the locus TP53, rising to 69% when regarding all 4 markers on 17p. Therefore, our data are in line with the assumption of additional tumour suppressor genes on 17p in HNSCC. A nuclear accumulation of p53 (51%) was independent from the antibody and the recognised epitope. At the first glance there was no correlation between overall p53 mutation (36%) and overexpression. However, it appeared that, with very few exceptions, only nonsense mutations did not lead to p53 overexpression, while missense mutations did. As overexpression of p53 was 15% more frequent than p53 mutations and only 35% of the tumours with p53 overexpression carried a p53 mutation, our data support the hypothesis of additional mechanisms of p53 overexpression. AI/LOH at the p53 locus in 83% of all tumours with a p53 mutation is in line with Knudson's theory of inactivation of tumour suppressor genes.

Malignant sinonasal tumours of epithelial origin are uncommon. The incidence is estimated to be less than 1:100,000/year (Robin et al., 1979); however, they generally tend to be aggressive, delicate to treat, and are frequently diagnosed only at an advanced stage. The histological typing of tumours of the upper respiratory tract and the ear by the World Health Organization specifies 19 different histological types of malignant epithelial tumours arising in the nasal cavity and paranasal sinuses (Shanmugaratnam, 1991). Sinonasal carcinomas, the most frequent subtype, can be differentiated histologically into the more frequent keratinizing squamous cell carcinomas and the very rare non-keratinizing, cylindrical cell, or transitional-type carcinomas. Non-keratinizing or transitional-type carcinomas accounted for only 8% of all sinonasal carcinomas in the study of Robin et al. (1979). A small percentage of transitional-type carcinomas may arise in preexisting transitional cell papilloma, as reported by Robin et al. (1979) and Svane-Knudsen et al. (1998). Knowledge on the underlying genetic changes in sinonasal malignant tumours of epidermal origin is in general very limited and there are no reports on molecular genetic changes in transitional-type carcinomas of the nasal cavity and the paranasal sinuses so far. We have analysed fresh frozen tumour specimens from 7 such patients, with their written and informed consent. In 3 tumour specimens (tumour 1: pT3N0M0 tumour from the right lateral nasal wall; tumour 2: pT2N0M0 tumour originating from the nasal septum; tumour 3: pTxN3M0 lymph node metastasis at the left mandibular angle) the histologic diagnosis was non-keratinizing, transitional-type carcinoma, or cylindrical cell carcinoma (ICD-O 8121/3). In tumour 3, the diagnosis was also based on immunohistochemical results with negative intra- and extracellular staining with periodic acid-Schiff (PAS), strong expression of cytokeratin 5, 10, and 11 and weak expression of cytokeratins 8 and 18. Tumour 4 was a pT3N0M0 keratinizing squamous cell carcinoma of the left maxillary sinus (ICD-O 8070/3); tumour 5 was a pT4N2aM0 adenocarcinoma of the left choana with deep infiltration of the skull base and the nasopharynx (ICD-O 8140/3); and tumour 6 was a pT4N0M0 adenoid cystic carcinoma of the left ethmoid sinus with infiltration of the orbit and the skull base (ICD-O 8200/3). Additionally, one transitional cell papilloma arising at the left lateral nasal wall (tumour 7, ICD-O 8121/0) was analysed. DNA from the tumour and from leukocytes was extracted using standard procedures. Loss of heterozygosity (LOH) was investigated at 15 genomic locations with high frequencies of LOH in many histological types of human cancer using polymorph microsatellite markers (Table I). Sequences of the primers were used as described by the Genome Data Base (www.gdb.org/). After amplification by polymerase chain reaction (PCR), the PCR products were analysed on 6% denaturating acrylamide gels, transferred to a nylon membrane and hybridised with 32P-labeled primer for 6 hr. Following LOH results, exons 1–3 of the von Hippel-Lindau disease gene (vHL), exon 3–10 of fragile histidine triad (FHIT), and exon 5–9 of p53 were investigated by single-strand conformational polymorphism (SSCP) analysis. The SSCP was performed at least twice with the addition of 5% and 10% glycerin to 6% non-denaturating polyacrylamide gels. The gels were run at 250 V, 50 mA, and 15 W for 4–5 hr and were constantly water cooled. The bands were visualised by routine silver staining. For double-strand sequencing analysis following SSCP analysis, an ABI Prism BigDye Terminator Cycle Sequencing Ready Reaction Kit was used (Applied Biosystems, Weiterstadt, Germany). Subsequent electrophoresis was performed using an ABI Prism 377 DNA Sequencer. Sequencing data were screened with the Analysis 1.1.1 software (Applied Biosystems) and analysed with the Sequencer 3.0 software (Gene Codes, Ann Arbor, MI). Immunochemical analyses were performed on frozen tissue to assess the expression of the putative tumour suppressors FHIT, vHL and p53. Monoclonal antibodies were G59–12 and DO-7 against p53, Ig32 against vHL (PharMingen, Hamburg, Germany) and ZR44 against FHIT (Zymed, San Francisco, CA). Biotinylated anti-mouse and anti-rabbit IgG as a secondary antibody and the avidin-streptavidin peroxidase protocol were used for staining. The expression was tested from at least 3 different tumour sites in 3 staining procedures for each protein and each tumour. A variety of frozen neoplastic and non-neoplastic tissues and cell lines were used as positive and negative controls. Tumour 1 exhibited LOH at 3p25.1 and 3p14.2 by using microsatellites D3S656 and D3S1234. Microsatellite D3S656 is close to the tumour-suppressor gene vHL and D3S1234 is located between exon 8 and 9 of the tumour-suppressor gene FHIT. This tumour also showed LOH at 8p23.22 (D8S552), whereas TP53 at 17p13.1, the location of the tumour-suppressor gene p53, retained heterozygosity. Tumour 2 also showed LOH at 3p25.1 with a homozygous constitution at 3p14.2. This tumour showed a heterozygous constitution at 17p13.1. Tumour 3 exhibited LOH solely at 17p13.1 using TP53 and exhibited a homozygous constitution at 3p25.1 and a heterozygous constitution at 3p14.2. Examplary results of the LOH analysis from tumours 1–3 are shown in Figure 1. Tumour 4 (a keratinizing squamous cell carcinoma arising in the maxillary sinus) also showed, like cases 1 and 2, LOH at 3p25.1 and additionally showed LOH for D5S82, which is located at 5q14–21. This region contains the tumour-suppressor gene APC. Tumour 5 (an adenocarcinoma) also showed LOH with marker D5S82 and additionally LOH at 4q25 (D4S407), 10q25–26 (D10S587) and 11p14–13 (D11S904). Tumour 6 (an adenoid cystic carcinoma) showed no LOH at any of the investigated locations, and, interestingly, neither did case 7, a sinonasal papilloma (Table I). Following the results of LOH analysis, with detected genomic alterations at 3p14.2, 8p23–22, and 17p13.1 in 1 and 3p25.1 in 2 cylindrical cell carcinomas, we analysed the coding regions of the tumour-suppressor genes vHL, FHIT and p53 for intragenic mutations by SSCP analysis and double-strand sequencing in case of suspect SSCP findings. None of the tumours 1–7 carried a mutation leading to a change in the amino acid sequence. In tumour 3, we detected a polymorphism in exon 8 of the FHIT gene, codon 98 [CAT(His)-CAC (His)]. In tumour 6 (an adenoid cystic carcinoma) we detected a silent mutation in exon 6, codon 213 [CGA(Arg)-CGC(Arg)] of p53. We also investigated the expression of the tumour-suppressor genes vHL, FHIT and p53 by immunohistochemistry. Every tumour specimen (1–7) showed strong expression of vHL. Particularly tumours 1, 2 and 4, with detectable LOH at the site of the tumour-suppressor gene vHL, showed no reduction of expression compared with all other tumour specimens and non-neoplastic nasal mucosa. However, there was only minimal intensity of FHIT staining in tumours 1 and 2, both showing LOH at chromosomal arm 3p. All other tumour specimens as well as non-neoplastic nasal mucosa expressed FHIT. Interestingly, none of the tumours (1–7) showed nuclear accumulation of p53, which is known to be an indicator for p53 alteration. Autoradiographs of microsatellites D3S656, D3S1234 and D8S552 (case 1), D3S656 (case 2) and TP53 (case 3) showing loss of one allele in the tumour. Left lane: nonneoplastic DNA from lymphocytes; right lane: DNA extracted from tumour tissue. Because of the presence of nonneoplastic tissue in the specimens, a faint band remains visible instead of complete loss. It would be of particular interest for the understanding of multistep transformation into malignancy to establish a genetic model of tumourigenesis in transitional type carcinoma, particularly when arising in its putative precursor, transitional type papilloma. However, the factors contributing to the genesis of sinonasal carcinomas as well as sinonasal papillomas are poorly understood. Viral cofactors in the etiology of sinonasal papilloma have been suggested (Macdonald et al., 1995; Mirza et al., 1998). The understanding of the role of the tumour-suppressor gene p53 in sinonasal papilloma is limited to 2 immunohistochemical studies of p53 expression in sinonasal papillomas (Mirza et al., 1998) and sinonasal papillomas compared with associated squamous cell carcinomas (Fang et al., 1998). In both studies, nuclear accumulation of p53 was detected in part of the papillomas as well as in squamous cell carcinomas: Mirza et al. (1998) detected p53-positive cells in 30% of sinonasal papillomas whereas the only cylindrical cell papilloma of their study showed weak p53 staining. Our understanding of genomic changes in sinonasal papillomas is restricted to the cytogenetic analysis of Jin et al. (1997) on 3 inverted nasal papillomas, in which all 3 short-term cultures exhibited cytogenetic abnormalities, and a detailed cytogenetic analysis of 6 short-term cultures derived from malignant tumours of the nasal cavity and the paranasal sinuses, including 3 squamous cell carcinomas (Jin et al., 1995). All investigated tumours showed complex chromosomal rearrangements. So far, our findings on LOH in malignant epithelial nasal tumours cannot be compared with LOH findings in large series of sinonasal papillomas; however, in our study, the transitional cell papilloma (case 7) showed no allelic loss or allelic imbalance at all, as compared with allelic imbalance at multiple loci demonstrated in cases 1 and 2, suggesting that a possible passage from transitional cell papilloma to malignant cylindrical cell, transitional-type carcinoma corresponds to a considerable gain in genomic instability. The inactivation of p53 appears to play no etiologic role in the 3 cylindrical cell carcinomas of our study and in neither of the additionally analysed malignancies, as only one tumour (case 3) showed LOH at 17p13.1 and none of the tumours carried a mutant p53 allele leading to a change in the amino acid sequence. Furthermore, none of the analysed specimens showed overexpression or nuclear accumulation of p53 protein, which would be indicative of p53 alteration, due to the short half-time of the wild-type protein. In our opinion, this is an important hint for a specific genetic pathway of epithelial neoplasms of the nasal cavity and the paranasal sinuses, because p53 alterations are among the most frequent and the earliest genetic changes in many histologic types of human cancer, including head and neck squamous cell carcinoma. Although 17p alterations were not frequent, and p53 alterations could not be detected at all, 3p alterations were present in case 1, as demonstrated by LOH at 3p14.2 and 3p25.1, and in case 2 with LOH at 3p25.1 and a homozygous constitution at 3p14.2. Furthermore, the keratinizing squamous cell carcinoma in our study also showed LOH at 3p25.1, making this chromosomal arm the most frequently altered region that we could identify. This chromosomal arm is affected by allelic imbalance in many types of human cancer, including oral cancer and nasopharyngeal cancer (Deng et al., 1998; Ishwad et al., 1996). Chromosomal arm 3p harbours the tumour-suppressor genes vHL at 3p25 and FHIT encompassing the FRA3B fragile region at 3p14.2. While vHL was not mutated in any of our specimens and the expression was uniformly strong in all specimens, FHIT expression was markedly decreased in cases 1 and 2, as assessed by immunohistochemistry. We did not detect any intragenic mutations within exon 3–10 of FHIT; however, this gene is more frequently altered by homozygous intragenic deletions rather than mutations (Druck et al., 1997). Even if we did not demonstrate homozygous deletions of FHIT, the reduced expression of this gene compared with non-neoplastic mucosa emerges as an indicator for the inactivation of FHIT in some cylindrical cell carcinomas. Yours sincerely, Karl Götte karl.goette@hno.ma.uni-heidelberg.de, Frank Riedel, Carsten Schäfer, Johannes Coy, Karl Hörmann

Vascular endothelial growth factor (VEGF) has potent angiogenic activity and has been identified in a wide variety of malignancies, including head and neck squamous cell carcinoma (HNSCC). The tumour-suppressor gene p53 has been thought to regulate VEGF. Cryostat sections of 33 head and neck squamous cell carcinomas (HNSCC) were immunostained for VEGF using a standard streptavidin-biotin complex procedure. To evaluate angiogenesis, microvascular density was counted by staining endothelial cells immunohistochemically using anti-vWF monoclonal antibody. The p53 gene status was analysed using a PCR-SSCP analysis and direct sequencing. VEGF positive staining was detected in 18 (55%) out of 33 tumours. VEGF immunoreactivity did not correlate with the main clinicopathological characteristics of the patients (localization, T-stage, N-status, histological grading). Statistical analysis gave a clear correlation between the tumour vascularity and the VEGF protein expression (p = 0.0036). VEGF negative tumours showed a lower mean number of microvessels per microscopic field (60.3 +/- 15.5) than VEGF positive tumours (79.6 +/- 22.9). P53 mutations were identified in 12 (36.4%) of 33 tumours. The association of p53 mutations and VEGF expression level was significant (0.027). The higher microvessel density in VEGF positive tumours supports the importance of VEGF for tumour angiogenesis in HNSCC. Our results support the hypothesis of a p53 regulation on the angiogenic process through a VEGF up-regulation.