Vergleich der publizierten Genexpressionsdaten beim Pankreaskarzinom

Art des Experiments: Meta-Analyse
Studienziel: Vergleich der publizierten Genexpressionsdaten beim Pankreaskarzinom
Software: Excel
Kriterien für die von Kanditatengenen: Differentielle Expression in mindestens zwei Experimenten verschiedener Arbeitsgruppen Ergebnisse: During the last years microarray analysis has become a standard technique to identify genes differentially expressed in various tumour entities. Up till now seven papers showing an aberrant gene expression profile in pancreatic cancer have been published. Comparing these data and our own results, we surveyed a total number of 548 differentially expressed genes. We found 86 genes described to be differentially expressed in more than one study. The majority of these genes (n=70) were found to be upregulated in pancreatic cancer, whereas 16 were revealed to be downregulated. Only a single gene, i.e. S100 calcium binding protein P, was described to be upregulated in five out of the eight studies. Four genes, among them integrin beta4, were found to be over expressed in four studies. Three studies corresponded in describing a group of nineteen genes as over expressed. Revealing the conformity in two studies a set of forty-six genes was upregulated. Comparing these recent microarray data with previous publications overall 9 genes, including Stratifin and CD55, have been implicated in pancreatic cancer before. These examples may support the validity of microarray-based expression profiling. In addition 14 of the upregulated genes in pancreatic cancer are known to be involved in other tumour entities. The remaining candidates have not been implicated in carcinogenesis yet. Striking discrepancies in these studies were restricted to only one gene, i.e. NAP1L1. There are several potential reasons for the overall rather low concordance of these studies: the type, histology and number of investigated samples differed considerably (i). Microdissection was applied in the study of our group only (ii). Different types of arrays and hybridisation technologies may cause different results (iii). There is no standardised procedure for statistical analysis and data mining (iv). Taken together, the results of expression profiling studies in pancreatic adenocarcinoma are not yet really comparable. This fosters the need of further verification steps investigating the differentially expressed candidate genes by independent methods such as real time RT-PCR. Furthermore, the validation must extend the efforts to protein level using immunohistochemistry or Western blotting procedures.