Prediction of colorectal cancer chemotherapy efficacy using machine learning applied to gene expression data.

Abstract

e15548 Background: FOLFOX and FOLFIRI chemotherapy are considered standard first-line treatment options for colorectal cancer (CRC). However, the criteria for selecting the appropriate treatments have not been thoroughly analyzed. Here, we aim to use machine learning and genetic profiles to identify precise multi-gene panels that can predict the response to 5-Fluorouracil-based chemotherapy in colorectal cancer patients. Methods: The genetic profiling data, including drug response profiles, were retrieved from the Gene Expression Omnibus (GEO) database. These datasets were used to train and validate machine learning models. Feature selection methods, such as least absolute shrinkage and selection operator (LASSO) and variable selection from random forests (varSelRF) and various algorithms, such as Random Forest and Support Vector Machines, were applied to develop predictive models. Functional enrichment and network analyses were performed using Ingenuity Pathway Analysis (IPA). Results: This study identified relevant gene signatures at two stages of colorectal cancer: primary and metastasis using two different chemotherapy regimens, FOLFOX and FOLFIRI. The predictive models achieved an average prediction accuracy of 93% in identifying drug response outcomes across multiple chemotherapy regimens. The application of the machine learning model suggested that 28.6% of patients who failed the treatment therapy they received would benefit from the alternative treatment. Conclusions: The developed machine learning models demonstrate potential for guiding clinicians in selecting the most effective treatment options based on individual genetic profiles. With additional clinical validation, this approach could lead to improvements in treatment outcomes for patients with CRC and other cancers.