Researchers Develop Diagnostic Tool to Identify Treatment of Certain Lung Cancers
Researchers at the Moffitt Cancer Center in Tampa, Florida have developed a non-invasive diagnostic tool to help doctors analyze tumor mutations and determine the best course of action for treatment. The research pertains to non-small cell lung cancer (NSCLC), the most common subtype of lung cancer as well as the leading cause of cancer-related deaths in the world.
There are several treatment strategies for NSCLC, including surgery, radiation therapy, and chemotherapy. Two of the most commonly used, however, are tyrosine kinase inhibitors and immune checkpoint inhibitors. The question of which treatment is better for certain patients is one that doctors have been struggling with for years. Doctors usually choose their strategy based on heterogeneous biomarkers, but in many cases, these biomarkers change during therapy and ultimately cause the treatment to be ineffective.
The new research, “Noninvasive decision support for NSCLC treatment using PET/CT radiomics,” published by Nature Communications, will help doctors determine which treatment method is best by analyzing patients’ tumors and biomarkers. Moreover, it uses a non-invasive diagnostic tool, which means that no invasive or potentially life-threatening procedures are required.
Deep-Learning Model Used
The Moffit research team employed a deep-learning model that uses positron emission tomography and computerized tomography radiomics to identify which NSCLC patients may be sensitive to tyrosine kinase inhibitor treatment (thereby rendering the treatment ineffective) and which patients may benefit from immune checkpoint inhibitor therapy.
The model uses a specific type of imaging called 18F-FDG PET/CT, which is essentially PET/CT scan imaging with the radiotracer 18F-Fluorodeoxyglucose, a type of sugar molecule. This type of imaging can help scientists identify abnormal glucose metabolism and characterize tumors accurately.
The way the model works is by showing high accuracy in epidermal growth factor receptor (EGFR) mutation status prediction. EGFR are small proteins that are found on the surface of all cells and proteins circulating in the blood, and they are responsible for the growth of certain lung cancers. In normal cells, the binding of EGFR and growth factors stimulates biological processes within the cell, which promotes strictly controlled growth. In cancer cells, EGFR is significantly overexpressed or constantly active, both of which lead to uncontrolled growth, or mutation.
Patients with an active EGFR mutation typically have a better response to tyrosine kinase inhibitor treatment, which is why identifying EGFR mutation status is so crucial. Drugs that block EGFR activity can slow cancer and perpetuate survival.
Moffitt Cancer Center Study Most Accurate Compared to Previous Studies
According to researchers, prior studies have shown that using non-invasive radiomics can predict EGFR mutation, but the research conducted at Moffitt Cancer Center yielded the most accurate results from all previous studies.
“Prior studies have utilized radiomics as a noninvasive approach to predict EGFR mutation,” said Wei Mu, PhD, first author of the study and postdoctoral fellow at the Cancer Physiology Department at Moffitt. “However, compared to other studies, our analysis yielded among the highest accuracy to predict EGFR and had many advantages, including training, validating, and testing the deep learning score with multiple cohorts from four institutions, which increased its generalizability.”
Moffitt researchers developed their deep learning model using retrospective data from NSCLC patients at the Shanghai Pulmonary Hospital and Fourth Hospital of Hebei Medical University in China.
Robert Gillies, PhD, the chair of the Cancer Physiology Department at Moffitt, hopes that the model they developed will serve as a basis for doctors to make clinical decisions. He explained: “We found that the EGFR deep learning score was positively associated with longer progression-free survival in patients treated with tyrosine kinase inhibitors, and negatively associated with durable clinical benefit as well as longer progression-free survival in patients being treated with immune checkpoint inhibitor immunotherapy.” He continued: “We would like to perform further studies but believe this model could serve as a clinical decision support tool for different treatments.”
The deep learning model can help scientists overcome the heterogeneous biomarkers that change over the course of therapy and thereby render certain treatments ineffective. This will help doctors identify the best course of treatment for patients with NSCLC, and it will all be done through a non-invasive diagnostic process. Some 5% of patients diagnosed with metastatic disease have a survival rate of five years. Researchers hope that their new model will contribute to a higher survival rate.