Although anti-cancer treatments have significantly advanced over the past decades, obstacles to accomplishing successful treatment still exist. The occurrence of treatment resistance is one of the... Show moreAlthough anti-cancer treatments have significantly advanced over the past decades, obstacles to accomplishing successful treatment still exist. The occurrence of treatment resistance is one of the major factors that limit the long-lasting efficacy of anti-cancer treatment. Additionally, substantial variability in pharmacokinetics (PK) / pharmacodynamics (PD) of anti-cancer drugs also challenges successful oncology treatment. Therefore, gaining knowledge of and ultimately better suppressing evolutionary resistance development during treatment, and applying personalized treatment are desired to improve anti-cancer treatment. In this thesis, we have applied quantitative modeling approaches to address these needs, aiming for improved treatment for oncology patients. Our work demonstrated that with the quantitative models, the evolutionary progression of tumors could be characterized and predicted, accounting for interactions among heterogeneous tumor cells and supported by mutant gene variants detected in circulating tumor DNA (ctDNA). In addition, we developed population PK /PD models which enabled quantitative description of the PK and PD of anti-cancer drugs and corresponding variabilities in real-world patients. The developed models have been further applied to support the identification of optimal treatment strategies and guide individualized treatment for oncology patients. Show less
Breast cancer is the most prevalent cancer in women. Early detection of this disease improves survival and therefore population screenings, based on mammography, are performed. However, the... Show moreBreast cancer is the most prevalent cancer in women. Early detection of this disease improves survival and therefore population screenings, based on mammography, are performed. However, the sensitivity of this screening modality is not optimal and new screening methods, such as blood tests, are being explored. Most of the analyses that aim for early detection focus on proteins in the bloodstream. In this study, the biomarker potential of total serum N-glycosylation analysis was explored with regard to detection of breast cancer. In an age-matched case-control setup serum protein N-glycan profiles from 145 breast cancer patients were compared to those from 171 healthy individuals. N-glycans were enzymatically released, chemically derivatized to preserve linkage-specificity of sialic acids and characterized by high resolution mass spectrometry. Logistic regression analysis was used to evaluate associations of specific N-glycan structures as well as N-glycosylation traits with breast cancer. In a case-control comparison three associations were found, namely a lower level of a two triantennary glycans and a higher level of one tetraantennary glycan in cancer patients. Of note, various other N-glycomic signatures that had previously been reported were not replicated in the current cohort. It was further evaluated whether the lack of replication of breast cancer N-glycomic signatures could be partly explained by the heterogenous character of the disease since the studies performed so far were based on cohorts that included diverging subtypes in different numbers. It was found that serum N-glycan profiles differed for the various cancer subtypes that were analyzed in this study. Show less
In this dissertation we developed a number of automatic methods for multi-modal data registration, mainly between mass spectrometry imaging, imaging microscopy, and the Allen Brain Atlas. We... Show moreIn this dissertation we developed a number of automatic methods for multi-modal data registration, mainly between mass spectrometry imaging, imaging microscopy, and the Allen Brain Atlas. We have shown the importance of these methods for performing large scale preclinical biomarker discovery investigations for neurological disorders. We have also proposed a data-driven approach to stratify patients’ tumor tissues into molecularly distinct tumor subpopulations and automatically identify those tumor subpopulations that drive patient outcome. Show less