Chromosomal aberrations in benign and malignant salivary gland myoepitheliomas

Salivary gland myoepithelial tumors are relatively uncommon tumors with an unpredictable clinical course. More knowledge about their genetic profiles is necessary to identify novel predictors of disease. In this study, we subjected 27 primary tumors (15 myoepitheliomas and 12 myoepithelial carcinomas) to genome-wide microarray-based comparative genomic hybridization (array CGH). We set out to delineate known chromosomal aberrations in more detail and to unravel chromosomal differences between benign myoepitheliomas and myoepithelial carcinomas. Patterns of DNA copy number aberrations were analyzed by unsupervised hierarchical cluster analysis. Both benign and malignant tumors revealed a limited amount of chromosomal alterations (median of 5 and 7.5 respectively). In both tumor groups, high frequency gains (≥20%) were found mainly at loci of growth factors and growth factor receptors (e.g. PDGF, FGF(R)s, and EGFR). In myoepitheliomas, high frequency losses (≥20%) were detected at regions of proto-cadherins. Cluster analysis of the array CGH data identified three clusters. Differential copy numbers on chromosome arm 8q and chromosome 17 set the clusters apart. Cluster 1 contained a mixture of the two phenotypes (n=10), cluster 2 included mostly benign tumors (n=10), and cluster 3 only contained carcinomas (n=7). Supervised analysis between malignant and benign tumors revealed a 36 Mbp-region at 8q being more frequently gained in malignant tumors (p=0.007, FDR=0.05). This is the first study investigating genomic differences between benign and malignant myoepithelial tumors of the salivary glands at a genomic level. Both unsupervised and supervised analysis of the genomic profiles revealed chromosome arm 8q to be involved in the malignant phenotype of salivary gland myoepitheliomas. FFPE material was used throughout the study. 15 primary myoepitheliomas and 12 malignant myoepitheliomas were analyzed on two different BAC platforms, both with 1 Mb spatial resolution. To investigate amplified regions in more detail, a 44k oligo platform was employed on a malignant myoepithelioma. After extraction, DNA quality was assessed by isothermal amplification (Buffart et al., 2007). Only DNAs which scored as excellent, good, or intermediate were used for array CGH analysis. Genomic DNA (600 ng) from both the test and reference sample was labeled and hybridizations were essentially performed according to Snijders et al. (Snijders et al., 2001) without the prehybridization step. Spot analysis and quality control for the BAC arrays were fully automated using Bluefuse 3.4 edition software (BlueGnome, Cambridge, UK). Spots with Quality flag <1 or Confidence value <0.3 were excluded. Log2ratios (tumor signal divided by normal reference signal) of spots that were not excluded after quality flagging and mapping were normalized to their block median value. For the array analysis, BAC clones were positioned along the genome using the March 2006 freeze of the UCSC database. For each clone the average of the triplicate spots was calculated and ratios were normalized by subtraction of the median value of all BAC clones on chromosomes 1-22. The sex chromosomes were discarded from the analysis, since all tumor samples were hybridized to reference DNA of the opposite gender.