Personalized treatment options for lung cancer patients have come a long way in the last two decades. Two major therapeutic strategies have emerged for patients with non-small cell lung cancer, the most common subtype of lung cancer, and the leading cause of cancer-related death worldwide: tyrosine kinase inhibitors and immune checks. Point inhibitor. However, choosing the right treatment for patients with non-small cell lung cancer is not always an easy decision, as biomarkers can change during treatment and the treatment may be ineffective. Researchers at the Moffitt Cancer Center are developing non-invasive and accurate methods for analyzing patient tumor mutations and biomarkers to determine the best course of treatment.
In a new article published in Nature CommunicationsThe research team found that deep learning models using positron-releasing tomography / computed tomography radiomics benefit from immune checkpoint inhibitor therapy as to which non-small cell lung cancer patients are sensitive to tyrosine kinase inhibitor therapy. Shows how to identify the patient to receive. This model uses PET / CT imaging with a radiotracer 18F-fluorodeoxyglucose, a type of sugar molecule. Imaging with 18F-FDG PET / CT helps identify sites of abnormal glucose metabolism and accurately characterize tumors.
This type of diagnostic imaging, 18F-FDG PET / CT, is widely used in determining the staging of patients with non-small cell lung cancer. The glucose radiotracer used is also known to be affected by EGFR activation and inflammation. EGFR, or epidermal growth factor receptor, is a common mutation found in patients with non-small cell lung cancer. Patients with active EGFR mutations respond better to tyrosine kinase inhibitor treatment, so the status of EGFR mutations may be a predictor of treatment. “
Dr. Matthew Shabas, associate member of the Faculty of Cancer Epidemiology
For this study, the Mofit team used retrospective data from non-small cell lung cancer patients at two facilities in China (Shanghai Lung Hospital and Hebei Medical University Hospital 4), based on 18F-FDG PET / CT. We have developed a deep learning model for. The model classifies EGFR mutational states by generating an EGFR deep learning score for each patient. The researchers created further validated the model using patient data from two additional institutions, Harbin Medical University’s Fourth Hospital and Morfit Cancer Center.
“Previous studies used radiomics as a non-invasive approach to predicting EGFR mutations,” said Dr. Wei Mu, Ph.D., lead author and postdoctoral fellow in the Department of Cancer Physiology. “But compared to other studies, our analysis was obtained with the highest accuracy for predicting EGFR, training, validation and testing of deep learning scores in multiple cohorts from four institutions. There are many advantages, such as, and its generalizability has increased. “
“EGFR deep learning scores are positively correlated with progression-free survival in patients treated with tyrosine kinase inhibitors and with the permanent clinical benefit of patients treated with immune checkpoint inhibitor immunotherapy. We found that there was a negative correlation with prolongation of progression-free survival, “Robert said. Gillies, Ph.D. , Chairman of the Department of Cancer Physiology. “I would like to do more research, but I believe this model will serve as a clinical decision-making tool for a variety of treatments.”
H. Remofit Cancer Center & Laboratory
Mu, W. , et al. (2020) Non-invasive decision support for NSCLC treatment using PET / CT radiomix. Nature Communications. doi.org/10.1038/s41467-020-19116-x.