Hollows, Robert John (2012)
M.Res. thesis, University of Birmingham.
Aberrant DNA methylation of CpG sites has been linked to the aetiology and pathogenesis of various malignancies including acute myeloid leukemia (AML). Decitabine is a drug which has been shown to reduce methylation levels, and is being increasingly explored as an agent for treating cancer. However, the determinants of demethylation caused by decitabine are not completely understood.
The purposes of this study were to investigate the determinants of the demethylation observed following treatment of AML samples with decitabine, and to explore whether these determinants could explain the variation in demethylation across two AML cell lines and eight primary cultures from AML patients.
The results showed considerable variation in the level of demethylation between samples. Within samples, CpG demethylation was found to vary according to CpG location, CpG density, proximity to a CpG island and pre-treatment methylation levels. Multivariate regression analysis showed that the principal determinant of demethylation at an individual CpG site was the pre-treatment methylation level. However, the analysis also showed that the determinants identified were in themselves insufficient to explain all of the variation in demethylation observed across study samples.
In recent years new "omic"-based technologies, such as microarrays, have been used to create many novel biomarker models for predicting outcomes in human cancer. However, only a fraction of these models have been put into actual clinical use. For models to have proven value, they must be shown to be generalisable to the wider population away from the data sets used to create them. To achieve this models must be properly validated. Various studies have considered how such validation should be performed.
In this study, a survey has been undertaken of 100 recent papers which have claimed to have validated "omic"-based biomarker models. The purpose of the study was to compare actual validation methodologies being used against best practice as set out in the literature. The results show that there are considerable deficiencies in the way that validation is undertaken, in particular with regard to sample sizes which are too small, inappropriate handling of data and over-reliance on validation methods which do not use genuinely independent data. Also, there is a disappointing shortage of studies undertaking independent validation of models constructed by other research teams.
In conclusion, more emphasis is required on the proper validation of biomarker models.
This unpublished thesis/dissertation is copyright of the author and/or third parties.
The intellectual property rights of the author or third parties in respect of this work are as defined by The Copyright Designs and Patents Act 1988 or as modified by any successor legislation. Any use made of information contained in this thesis/dissertation must be in accordance with that legislation and must be properly acknowledged.
Further distribution or reproduction in any format is prohibited without the permission of the copyright holder.
Repository Staff Only: item control page