Posted: Friday, January 27, 2023
A novel epithelial-mesenchymal transition (EMT)-related gene signature has been identified and may be used to predict outcomes and therapeutic response in patients with bladder cancer, according to an article published in the journal Disease Markers. “The complex etiological variables and high heterogeneity of bladder cancer make prognostic prediction challenging. We aimed to develop a robust and promising gene signature using advanced machine learning methods for predicting the prognosis and therapy responses of patients with bladder cancer,” said authors Xiaopeng Hu, PhD, and colleagues, of Capital Medical University, Beijing, China.
The retrospective study used data associated with 779 patients with bladder cancer, acquired from The Cancer Genome Atlas and the Gene Expression Omnibus studies GSE13507 and GSE32894. These data included samples of bladder cancer and adjacent normal samples. The authors implemented the single-sample gene-set enrichment analysis algorithm and univariable Cox regression to identify risk factors associated with bladder cancer. Then, using machine learning methods with survival and differential gene expression, the group was able to identify nine EMT-related genes, three of which were identified as risk factors. The model was trained on the 393 patients from The Cancer Genome Atlas while using the two cohorts from the Gene Expression Omnibus to perform separate validations.
The model revealed that high-risk patients were more likely to have a poorer prognosis than those at low risk. To further understand the pathogenic mechanism of bladder cancer, the authors found that CD8-positive T-cell, Tregs, and M2 macrophage concentrations were high within the tumor microenvironment in patients at high risk. Furthermore, the model was able to predict that high-risk patients were more responsive to some chemotherapy drugs, whereas low-risk patients might respond better to immunotherapy.
Disclosure: The study authors reported no conflicts of interest.