Application and development of computational approaches to optimize treatment of malignancies using routine clinical data

Resumo

The use of personalised medicine in oncology is gaining recognition as a way of enabling individualised tailored-treatment based on patient’s genetic signatures and clinical characteristics. The characterisation of drug exposure and the way it affects to the dynamics of the underlying disease under study (either through tumour size changes or biomarker dynamics), is key to support individualised disease monitoring and therapeutic strategies. The field of pharmacometrics is a potentially useful discipline which focuses on obtaining quantitative mathematical and statistical models of the different physiological processes from drug administration to measurement of drug exposure, disease dynamics (tumour size, biomarker response), and ultimately clinical outcome. In this thesis we will present examples of the use of different ways of characterising disease dynamics and pharmacometric tools to facilitate personalised medicine. These examples include applications to enable individualised medicine in routine clinical practice and to support model-based drug development. The first section of the thesis, the Introduction, gives a general overview about the most relevant phamacometric concepts, and about the different methodologies that will be explored in the different chapters of the thesis, with special focus on the oncology area. Chapter 1 reviews the different drug exposure characterisation approaches that can be applied to the oncology field, providing guidance on the different use of the methods depending on the data available. Chapter 2 shows an example of the characterisation of the pharmacometric properties of two oncologic drugs, using data from different sources, and characterising and quantifying differences on the drug exposure between individuals. Chapter 3 presents a pharmacometric model describing the dynamics of tumour size in gastric cancer patients in a categorical manner. In this work, the prediction of response to a neoadjuvant approach is shown, including Markovian features on it. In addition, a time to event model of the progression-free and overall survival is shown. Chapter 4 proposes a pharmacometric framework to describe the longitudinal dynamics of tumour size in breast cancer patients, and to quantify the effect of neoadjuvant chemotherapy and immunotherapy on these patients. Additionally, the impact of tumour response on overall survival is also presented. Chapter 5 describes a PKPD model, linking exposure to Busulfan and Treosulfan, and neutrophil dynamics as markers of disease progression, in paediatric patients suffering from a Haematopoietic Stem Cell Transplant (HSCT). Therefore, the contribution of the HSCT to neutrophil dynamics is also studied. In addition, Busulfan and Treosulfan dosing schedule is evaluated. Chapter 6 presents an optimal design exercise performed based on the pharmacokinetic model developed on Chapter 5. In this chapter, optimal sample times are obtained in order to improve therapeutic drug monitoring and thus, improve personalised medicine in children. Finally, the General Discussion highlights the most relevant aspects of the six chapters and focuses on future perspectives of the work presented, and the Conclusions summarise the main findings of this thesis.