Proteomic studies in cancer : statistical methods, experimental insights, and the challenges of dual proteomics profiles

Cancer is fundamentally driven by genetic alterations that lead to changes in transcription and translation, which can be detected through Omics techniques. Proteomics, which provides information resulting from gene translation, elucidates the impact of genetic alterations on cellular functions and biological processes. However, proteomics studies typically suffer from low reproducibility due to the inherent variability of the proteome in organisms itself and external factors such as technical limitations and poor experimental designs. Consequently, relatively fewer published methods or pipelines are available for analysing proteomics data compared to other Omics fields, such as genomics. This thesis is motivated by two datasets from two separate cancer proteomics studies, each generating dual proteome profiles per sample. The first study involves linking proteome profiles of cancer patients with their responses to subsequent treatments, aiming to discover proteins predictive of treatment outcomes. Additionally, technological issues of proteomics that can introduce unwanted variability are addressed using the data from this study. The second study investigates an alternative source for studying cancer antigens in body fluids, which are more accessible than tumour cells, by comparing peptidome profiles derived from both sources. The conclusions and key contributions of this thesis are drawn from several key findings: (i) In pursuit of improving treatment strategy for oesophageal cancer patients, a proteomic study revealed proteins and protein complexes in patients’ plasma that may serve as predictors of treatment response. This suggests the potential to use simple blood samples to predict treatment outcomes in advance, allowing for more informed decisions on whether to proceed with treatments that could be toxic. (ii) In the field of blood-based proteomics, this thesis contributes by providing a comprehensive list of albumin-interacting proteins whose identification and quantification may be compromised due to albumin depletion, which is considered as a standard proteomics procedure. This guide serves as a valuable resource for future blood-based proteomics studies; helping for inform study design and explain potential resource of variation; (iii) Methodologically, this thesis provides an experimental design strategy by exploring the feasibility of integrating two proteomics workflows. It addresses methodological challenges and presents solutions for more reliable and interpretable results in proteomics studies. (iv) A diagnostic tool was developed by linking proteomics data with gene-level information, addressing gaps in proteomics experimental design and improving the accuracy of diagnostic approaches in cancer research. (v) Finally, with the goal of finding tumour antigens for potential use in immunotherapy for cancer treatment, this thesis investigated the feasibility of using alternative sources of tumour antigens, rather than relying solely on tumour biopsy tissues. The result demonstrates population-wide differences in peptidome profiles of the two cellular sources of tumour antigens, challenging the currently published belief and offering new insights into tumour antigen discovery.