A team led by researchers at Cedars-Sinai University of Health Sciences has developed a faster and cheaper way to determine which genes are expressed in cancerous tumors. The AI-based tools they describe in Cell could make personalized cancer treatment available to more patients.
The new tool, called Path2Space, predicts gene expression across tumor areas based on digital images of biopsy slides containing thin sections of tumor tissue that can be examined under a microscope.
Because tumors do not have the same composition and gene expression throughout, Path2Space predicts so-called “spatial” gene expression and extrapolates it at many different points within the tumor. This process takes only minutes and is significantly less expensive than traditional spatial gene expression profiling, which typically takes weeks and costs thousands of dollars.
This tool makes two major contributions. This will allow us and others to study larger datasets and understand the spatial structure of tumors. But what really motivates me is that if we successfully validate our tools in clinical trials, we have the potential to improve cancer treatment for patients. ”
Dr. Eitan Ruppin, Deputy Director, Cedars-Sinai Translational Research Institute, Lead Study Author
The researchers “trained” Path2Space using data from a large group of breast cancer patients for whom both biopsy slides and spatial sequences were available. We then tested the tool on three additional patient datasets to validate its performance.
“For each sample, we looked at the actual measured gene expression and compared it to the tool’s predictions,” said study co-lead author Dr. Eldad Shulman, a research scientist at the National Cancer Institute who will soon join Ruppin’s lab as an investigator. “For each sample, we predicted the spatial expression of nearly 5,000 genes, and the predictions matched well with the expression measured across all three patient groups.”
Path2Space is designed to help scientists discover new biomarkers that can guide treatment decisions and identify patients at high risk of poor outcomes.
“This tool looks at features within the tumor, such as whether a gene is expressed in some regions of the tumor and not in others,” said Emma Campagnolo, co-first author of the study and a researcher in the Ruppin lab. “We discovered specific spatial patterns of gene activity in tumors that predict how patients will respond to treatment.”
Schulman said spatial biomarkers are difficult to identify because the high cost of spatial profiling through traditional methods means that very little data is available.
“Before we developed Path2Space, the largest cohort we could find to study the spatial organization of the tumor environment was approximately 30 patients,” said Schulman. “This tool allows us to study slides from thousands of patients.
Path2Space harnesses the potential of spatial biology in ways not possible before. ”
Path2Space could be applied to other types of cancer if trained on the right data, Campagnolo said, and the lab is finalizing studies to apply Path2Space to head and neck cancers. The team is also working on improving the accuracy of the tool. We are currently studying groups of 10 to 20 cells together, with the goal of eventually being able to evaluate individual cells.
“With the help of our clinical collaborators, we next hope to introduce Path2Space into clinical trials,” Ruppin said. “This is an exciting development in a growing field and will need to be tested carefully. However, we look forward to making significant contributions to science and patient care.”
Dr. Robert Figlin, interim director of Cedars-Sinai Cancer, said translational research is a hallmark of the institution.
“Developing tools that apply cutting-edge science to patient care is the best way to serve patients and improve cancer treatment worldwide,” Figlin said.
