Raffaele Argiento (Università di Bergamo and Collegio Carlo Alberto) - Principal Investigator
Francesco Stingo (Università di Firenze)
Matteo Ramazzotti (Università di Firenze)
Fondo di Beneficienza di Intesa San Paolo
March 2022 – February 2023
Cancer is a heterogeneous disease. Patients are characterized by a unique and complex genetic profile that will determine the etiology of the disease, and will affect how patients are managed in the clinic. Tumors that are apparently similar based on the tissue analysis are usually treated similarly in the clinic, although these tumors may be very different in terms of their genetic profile; with no surprise these apparently similar tumors may respond differently to the same therapy. Consequently, cancer treatment strategies have to take into account the genetic profile whose knowledge may allow clinicians to devise personalized treatment strategies that account for the unique features of each patient. This project CluB-PMx2 – aims at the development of novel statistical and data science methods for the selection of the optimal treatment based on the unique genetic profile of the patients (precision medicine). Given the complexity of the genetic data, the highly non-linear treatment effects and interactions, and the substantial uncertainty of any estimate of this type of models, we propose to develop sophisticated and flexible statistical models that can overcome the limitations of the state-of-the-art approaches. Specifically, we will develop a novel type of Bayesian clustering based on non-parametric techniques (Partition Models with covariates). The advantages of the proposed approach are the following: a) automatically identify similar genetic profiles, b) identify for each patient which one among a set of treatments is the one that will likely lead to the best therapeutic response c) determine the statistical relevance of the resulting inference.
CluB-PMx2 is a project that connects many researchers in statistics and oncology from Università Cattolica Milan, University of Florence and Policlinico Hospital in Naples. Genetic profiles are currently used in the clinic to suggest whether chemotherapy or hormonal treatment is the most appropriate treatment strategy for breast cancer patients.
CluB-PMx2 not only aims at developing new statistical methods for supporting precision medicine but also at promoting positive collaboration between practitioners/clinicians and the academic worlds. The data that can be collected with modern genomic technologies are characterized by original features: they have a complex structure, they are many and produced at hight velocity (these are peculiar characteristics of big data). To analyze these data new methods and new algorithms are needed, statistics play a fundamental role in this context. However, the new techniques must be discussed and improved with the support of clinicians who have a thorough knowledge of the application aspects of the problems under study.
From a methodological point of view, the target of CluB-PMx2 is to improve the ability to predict the response to different therapies of breast cancer’s patients. The idea is to fully exploit the information present in the predictive genetic signature for treatment selection by combining predictive and prognostic markers in a single flexible (non-linear) model that automatically cluster patients. The conjecture is that the use of nonparametric Bayesian techniques, although could slightly increasing the methodological complexity and the computational cost of the statistical procedure, will be very effective in predicting the response to a specific treatment of breast cancer patients. More generally we expect that our model can be applied to the study of other tumoral pathologies, and also to different pathologies such as cardiac pathologies.
Following the general philosophy of precision medicine, CluB-PMx2 aspire to provide new methodological tools for selecting the most effective treatment for groups of patients suffering from breast cancer. The final goal is to prescribe the best therapeutic results, avoiding unnecessary treatments. Although all breast cancer treatments present risks and side effects, avoiding unnecessary treatments will reduce some of these risks improving health-related quality of life of patients.