Abstract:
The identification of compounds that may be used as tumour markers has been in the focus of research for several years. Most of the established tumour markers show either a low sensitivity or specificity and are elevated significantly in only a few of the investigated cases. Other markers are suitable for surveying therapy, but not for cancer diagnosis. Especially in case of breast cancer, the diagnosis in an early state is difficult.
Modified nucleosides are end products of RNA metabolism and are excreted in cancer patients' urine in higher concentrations. This was observed in several studies and different types of cancer. Modified nucleosides could therefore be an alternative as tumour markers.
In this work, modified and normal nucleosides from biological fluids were characterized and identified using different mass spectrometric methods. These methods were applied for analysing urine samples and cell culture supernatants.
For semi-preparative separation of the nucleosides, an HPLC method was developed. The fractions were examined using MALDI-TOF-MS and nano-ESI-MS. A MALDI-TOF-MS-method was developed for measuring with high mass accuracy. Based on the accurate mass, a sum formula may be calculated and a data base search performed. With this method, eight unknown compounds in urine could be identified.
Furthermore, an auto-LC-MS3-method with an Ion Trap mass spectrometer was developed for structural information on unknown compounds. With this method, 38 ribosyl derivatives could be detected in urine samples of breast cancer patients. 16 of these nucleosides had been identified before; six were identified by comparison of retention time and fragmentation with standard compounds. For further 11 compounds, structure proposals were made based on the fragmentation and sum formula.
The auto-LC-MS3-method was also applied on cell culture supernatants and microsomal incubations. Supernatants of MCF7 breast cancer cells were compared to those of MCF10A epithelial cells.
Finally, all identified nucleosides and further potential ribosyl derivatives were examined for their potential as tumour markers. For classification of urine samples of breast cancer patients and healthy volunteers based on the relative concentrations of 32 compounds, a support vector machine algorithm was applied.