The dissertation is dedicated to integrating the drug metformin into standard neoadjuvant chemotherapy (NACT) for breast cancer (BC) and demonstrating its modifying antineoplastic action by studying changes in clinical and morphological parameters. This research also included determining the role of morphological markers and their dependence on the use of the modifier, studying the connection between the development of side effects and adverse events, treatment options for these effects, and creating a mathematical model capable of predicting therapeutic damage outcomes to the tumor.
1. The first stage included a theoretical analysis of scientific developments and an evaluation of the feasibility of implementing modifiers, specifically metformin. This involved an analytical approach to selecting the modifier, assessing the prospects of execution, and monitoring the study in a comprehensive analysis of clinical-morphological, radiological, and laboratory studies and the treatment of involved patients.
2. The second stage involved treating patients divided into three groups with different metformin dosing regimens. Weekly analysis of blood laboratory indicators, clinical tolerability of systemic antitumor treatment, and results of pathohistological and immunohistochemical studies were conducted.
3. The third stage included data collection, database creation on the effectiveness of incorporating metformin into standard preoperative cytostatic treatment, statistical analysis, mathematical model creation, including integrating machine learning and creating a random forest algorithm, publication of obtained results, and writing the dissertation manuscript.
The scientific novelty of the obtained results lies in the following. The conducted study is the first Ukrainian research on the integration of metformin as a modifier in standard neoadjuvant chemotherapy protocols for breast cancer treatment. For the first time, an algorithm for the use of metformin in neoadjuvant chemotherapy for breast cancer has been developed. For the first time, the therapeutic damage to tumors in patients receiving neoadjuvant treatment with and without metformin has been compared. For the first time, adverse clinical events during the combination of chemotherapy with metformin have been evaluated. For the first time in routine practice, immunohistochemical determination of Cyclin D1 and androgen receptors was used to predict the effectiveness of neoadjuvant chemotherapy. The scientific understanding of the role of clinical-radiological and pathomorphological responses of breast tumors to neoadjuvant chemotherapy, particularly the possibility of performing breast-conserving surgeries, has been expanded. The obtained results have demonstrated the feasibility of using metformin in neoadjuvant chemotherapy for breast cancer. For the first time, a calculator has been created to adapt the assessment of therapeutic tumor damage according to G.O. Lavnikova to the universally accepted international classification "Residual Cancer Burden," allowing for comparison of treatment results with data from most international studies. For the first time, a mathematical model and a machine learning model (random forest algorithm) have been created, based on primary clinical and morphological data, to predict the achievement of complete morphological response, considering the impact of metformin in neoadjuvant chemotherapy for breast cancer. Integrating mathematical models and machine learning methods in clinical practice has further developed, enabling a more personalized approach to breast cancer treatment. The scientific understanding of therapeutic approaches that improve outcomes and the quality of life for patients has been expanded, which is the primary goal of modern oncology.
The work aimed to improve the outcomes of personalized methods of systemic neoadjuvant treatment for breast cancer by adding response modifiers (metformin) to therapy and identifying markers that can be used to predict and assess the quality of this therapy through the creation of a mathematical model.
