Back to search

BIOTEK2021-Bioteknologi for verdiskaping

ERA-NET: Resistance under treatment in breast cancer (RESCUER)

Awarded: NOK 6.3 mill.

Personalised cancer therapy means to find for each patient the best therapy. Why is this difficult, when we have up to 60 approved therapies and drugs against breast cancer? The reason is that breast cancer is not one diseases, but a collection of similar but at the same time fundamentally different diseases. In fact the heterogeneity is extreme, because it is the characteristics of the cancer cells (and of the cells surrounding the cancer in the micro-environment) that determines the diseases subtype. The characteristics of the cells is determined at molecular level, but their genes, the way the genes function, their mutations and the way they collaborate in pathways. Therefore, in order to determine the best therapy for a patient, we will depend on understanding how the gene signature of the individual cancer modifies the efficacy of a therapy. In addition, the characteristics of cells is changing in time, which can be short, for example during a 3 months therapy, or long term, over many years. Resistance to treatment is also occurring. These are cells that for example do not die if exposed to a drug. In this first year of the project we have worked with two clinical trials, which are both concluded, and have enrolled breast cancer patients. These two trials (called NeoAva and Neoletexe) have about 100 patients each, who have been treated with specific drugs during a 12 weeks period. What is special with these two trials, is that quite a lot of data have been collected about the patients and their response to treatment. We have prepared the data, so that we have now access to the whole cohort of patients. More importantly, we have develop further our model published two years ago, which allows to simulate the way the tumour reacts to a chemotherapy, so that we are now able to simulate a large number of cells together, indeed as many as millions of cells, which are captured in a typical biopsy. We have been able to do this by optimising our simulation algorithm, so that it can now run in one day, to reproduce the treatment effect of these millions of cells over the 12 weeks treatment. We have also started to discuss how the model parameters depend on gene expressions. One can think as if each patient has her own personal parameter values, which will depend on the personal gene signature. We have also worked on another Spanish clinical trial, called CORALLEEN, where 50 patients have been treated with a combination of two targeted drugs, which have the potential of replacing chemotherapy for a subset of breast cancer patients. We have developed a mathematical model capturing the most important interactions of these drugs with the cancer cells on an intracellular level. We investigated the predictive power of this model to forecast the patient-specific response to the drug treatment and showed that it could accurately predict patient outcome after treatment. Recent advances in single-cell RNA sequencing (scRNA-seq) have unveiled diverse cell phenotypes within breast tumors. However, understanding their impact on tumor biology and treatment response is limited due to sample scarcity and tumor complexity. To address this, we estimated cell fractions from bulk expression samples of over 2000 Neoadjuvant Chemotherapy (NAC) patients. By employing explainable Machine Learning (XML) and different spatial omics data we explored cell phenotype associations with NAC response in various breast tumor subtypes.

Breast Cancer (BC) is the first cause of cancer-related death in women worldwide. Breast cancer is classified into well-recognized molecular subtypes. Despite solid pre-clinical evidence, only some patients benefit from administering drug combinations, an indication that patient and tumor heterogeneity is still present in the current stratification. Out of the numerous possible combinations of approved drugs, only a few have been actually tried, and the choice of tested combinations has been to some degree arbitrary. This proposal seeks to develop new approaches and identify mechanisms of treatment resistance at systems level, exploring how the effectiveness of specific targeted therapies applied in different clinical trials is affected by patient- and tumor-specific conditions. For this purpose, the project will gather and integrate longitudinal multidimensional data from ongoing clinical trials and newly generated -omics using systems approaches, which combine sub-cellular/cellular and/or organ level in silico models and network analysis to build computational frameworks able to discover molecular signatures of resistance and predict patient response to combinatorial therapies. We aim to identify the physiological characteristics of non-responders vs. responders from existing and newly generated multi-omic data and biological samples from in vivo and ex vivo clinical studies of specific subtypes of BC patients treated with combination therapy. This new knowledge will be used to investigate the curative potential of new personalized drugs combinations. The overreaching goal is to develop computer “xenograft model” as a cost-efficient and better alternative in terms of ethics, availability to everyone, and animal use. The framework will include optimization algorithms to identify combinations of approved drugs with a high probability to work on individual or thin strata of patients.

Publications from Cristin

No publications found

No publications found

Funding scheme:

BIOTEK2021-Bioteknologi for verdiskaping