INNOVATIVE BIOSTATISTICAL AND BIOINFORMATIC APPROACHES IN THE ANALYSIS OF BREAST CANCER: COMPETING RISK SURVIVAL ANALYSIS THROUGH PSEUDO-VALUES AND COMPREHENSIVE EVALUATION OF METHODS FOR THE TUMOR MICROENVIRONMENT DISSECTION AVAILABLE AT THE PRESENT DAY.
Tesi di Dottorato
Data di Pubblicazione:
2020
Citazione:
INNOVATIVE BIOSTATISTICAL AND BIOINFORMATIC APPROACHES IN THE ANALYSIS OF BREAST CANCER: COMPETING RISK SURVIVAL ANALYSIS THROUGH PSEUDO-VALUES AND COMPREHENSIVE EVALUATION OF METHODS FOR THE TUMOR MICROENVIRONMENT DISSECTION AVAILABLE AT THE PRESENT DAY / D. De Bortoli ; co-supervisor: E. BIGANZOLI, P. BORACCHI ; supervisior: C. DESMEDT. DIPARTIMENTO DI SCIENZE CLINICHE E DI COMUNITA', 2020 Jan 09. 32. ciclo, Anno Accademico 2019. [10.13130/de-bortoli-davide_phd2020-01-09].
Abstract:
Since my personal original background was quite distant from the statistical bioinformatic approaches for data analysis, having a master degree in Sanitary Biotechnology and Molecular Medicine, my PhD fellowship was spent in building my skills in this field while studying and trying to contribute to the development of biostatistical and bioinformatic approaches to be applied in clinic, with a special focus on oncology, in the optic to contribute to the field of personalized medicine. Personalized medicine is indeed the ultimate goal for life sciences, particularly for oncology, and, in my opinion, a key aspect of the future wellness of humanity. Personalized medicine is the idea of developing the ability to identify the best therapeutic strategy for each unique person and its efficacy relies on having accurate diagnostic tests that identify patients who can benefit from targeted therapies. A striking example consists in the determination of the overexpression of the human epidermal growth factor receptor type 2 (HER2) in the routinely diagnosis of Breast Cancer (BC). HER2 is indeed associated with a worse prognosis but also predicts a better response to the medication trastuzumab; a test for HER2 was approved along with the drug (as a “companion diagnostic”) so that clinicians can better target patients' treatment. My thesis is composed by the description of the two projects that have mainly characterized my fellowship. Both projects rely on breast cancer (BC) and the objective of understanding the effects of chronic low inflammation, which has been studied in my projects as the leucocyte infiltration and the body mass index. The focus on BC derives from a practical aspect and an epidemiological aspect. The practical aspect consists on the fact that my group is part of a European research group, led by Christine Desmedt from Belgium, which allowed me to obtain unique data and to interact with experts of BC and bioinformatics from different countries. The epidemiological aspect is represented by the fact that breast cancer is actually a hot topic, being the second most common cancer worldwide and the first among women, but still open to investigations, since the complexity and variability of BC, reflected both at histopathological and molecular level, have proven challenging to classify and therefore to effectively treat to the present day. The first project presented, the tumor microenvironment (TME) dissection project, occupied the first part of my fellowship and was focused on the managing of an enormous quantity of data in order to compare different tools and approaches used to analyze breast cancer. This project consisted in a big European collaboration which tried to establish the reliability of bioinformatic tools in retrieving the TME composition by analyzing bulk transcriptome and methylome and comparing the obtained results to standard approaches, as the pathologist evaluation, and emerging methods, as digital image analysis. This project led to the preparation of a paper, which is currently under submission, under the supervision of Christine Desmedt, the leader of this breast cancer research group, and Elia Biganzoli, my supervisor and member of the cited group. The second project presented, the competing risk analysis through pseudo-values project, which characterized the third year of my PhD, is more focused on the statistical aspects of clinical data analysis and represent the arrival point of my studies of statistical methodology. The project consisted in the exploration of a forefront approach to the analysis of survival data based on pseudo-values, which has the desirable feature to generate measures with a clear and direct interpretation at a clinical level, becoming an invaluable tool for clinical decision m
Tipologia IRIS:
Tesi di dottorato
Keywords:
Deconvolution; Survival; Breast; Cancer; Microenvirorment; Pseudo-values
Elenco autori:
D. DE BORTOLI
Link alla scheda completa: