An integrated proteomic and bioinformatic analysis for the diagnosis and prognosis of cancer.

Parkinson, E., 2006. An integrated proteomic and bioinformatic analysis for the diagnosis and prognosis of cancer. PhD, Nottingham Trent University.

10183131.pdf - Published version

Download (44MB) | Preview


The advent of proteomics and high-throughput technologies has allowed scientists to derive protein expression patterns of potential use in predictive medicine. The application of bioinformatics to analyse complex data makes it possible to identify important protein biomarkers. These biomarkers may have predictive capability to determine, for example, the presence and progression of disease and how an individual patient might respond to therapy.

Mass spectrometry (MS) has increasingly become the method of choice for the analysis of complex samples and new MS systems have been developed that can rapidly profile and generate proteomic 'fingerprints' from tissue and body fluids. In particular, MALDI mass spectrometry coupled with Ciphergen chip technology (SELDI MS) has been widely used to identify discriminatory patterns to distinguish patients at different clinical stages of disease, for example, in ovarian, prostate, colon and breast cancer. All of these studies incorporate the use of computer algorithms to mine the proteomic data obtained from the mass spectra, allowing large cohorts of samples to be included into the analysis.

The aim of this study was to introduce the use of MS and bioinformatics to analyse the cancer proteome, in particular melanoma and breast cancer and to investigate the information obtained from profiling cell lines, tissue and serum samples, as well as evaluating the type of analytical methods currently available.

The methods used in this study for sample preparation and analysis demonstrate that good quality proteomic data from cell lines, tissue and serum can be obtained and that it is possible to generate discriminatory protein profiles that correlate with clinical outcomes when analysed using Artificial Neural Networks (ANNs). Through the analysis of the proteome of melanoma cell lines, it is possible to classify samples according to the presence of specific genetic mutations, the site of the tumour sample from which the cell line was derived, as well as the overall survival of a patient. Comparison of melanoma cell line proteomes and their tumour tissue of origin revealed that both sample types were able to provide discriminating patterns that correlated to clinical outcomes. This finding has significance for future proteomic-based biomarker discovery research where it is possible to use cell lines in place of "precious" tumour tissue for the identification of clinically relevant biomarkers. The presence of a basal phenotype, which signifies the aggressive nature of breast cancer, can be identified from the proteomic profiling of patients' breast cancer tissue. The analysis of melanoma patient serum was investigated and patterns that predicted the stage of disease, as well as disease progression, were identified, using SELDI MS and ANNs. These results demonstrate that it is possible to obtain clinically valid information from the proteome of samples derived from melanoma and breast cancer patients through the use of SELDI MS and ANN analysis.

Although SELDI MS has proven useful in generating protein profiles that can be used for identifying patients with different clinical outcomes, this technology has limitations. One aspect of the study was to determine if similar, or more accurate, discriminatory analysis could be achieved using higher resolution and higher sensitivity MALDI instrumentation. A set of melanoma cell line samples were subjected to SELDI MS and MALDI MS analysis and the data from both methods were analysed in the same way by ANNs. Slightly different sample preparation methods were used prior to MS analysis, thus the spectra obtained by SELDI MS and MALDI MS was dissimilar; the data revealed that MALDI MS did not improve upon the accuracy of classifying samples.

The work presented demonstrates a proof-of-principle of the different types of information that can be obtained from samples derived from melanoma and breast cancer patients. It has also been revealed that the analysis of MS spectra by ANNs can be used for predicting blind datasets which is not necessarily dependent on the MS method used; however, this is likely to have significant implications for biomarker identification as the different methods used will reveal different disease-associated proteins.

Item Type: Thesis
Creators: Parkinson, E.
Date: 2006
ISBN: 9781369314120
Divisions: Schools > School of Science and Technology
Record created by: Jeremy Silvester
Date Added: 25 Sep 2020 10:08
Last Modified: 20 Jul 2023 10:31

Actions (login required)

Edit View Edit View


Views per month over past year


Downloads per month over past year