A team of researchers at the University of Wisconsin-Madison has combined genomics and machine learning to develop an accessible test that can detect cancer early.
For many types of cancer, early detection can lead to better outcomes for patients. Scientists are developing new blood tests to analyze DNA to aid in early detection, but these new technologies have limitations such as cost and sensitivity.
Muhammad Murtaza, Professor of Surgery, UW School of Medicine and Public Health University of Wisconsin-Madison
in the investigation Published in Science Translational Medicine this week Researchers, led by Muhammad Murtaza, a professor of surgery at the University of Washington School of Medicine and Public Health, used machine learning models to examine DNA fragments from cancer cells in plasma. Using readily available experimental materials, the technique detected cancer early in most of the samples they studied.
“We are very excited to discover that early detection and monitoring of multiple cancer types may be feasible using such a cost-effective approach,” Murtaza said. said.
This approach relies on analyzing fragments of cell-free DNA. Such fragments are common in plasma, the liquid portion of blood. Fragments of genetic material usually come from blood cells that die as part of the body’s natural processes, but they can also be shed by cancer cells.
The researchers found that DNA fragments from cancer cells may differ from healthy cell fragments in terms of where the DNA strand breaks and which nucleotides (the building blocks of DNA) surround the breakpoint. I hypothesized that there is.
Using a technique they dubbed GALYFRE (from genome-wide analysis of fragment ends), the team analyzed cell-free DNA from 521 samples and added additional samples from healthy individuals and patients with 11 different cancer types. We sequenced data from 2,147 samples.
From these analyses, they developed a scale that reflects the proportion of cancer-derived DNA molecules present in a sample. They called this information-weighted portion the anomalous fragment.
They used this measurement, along with information about the DNA sequences around the fragment breakpoints, to develop a machine learning model that compares DNA fragments from healthy cells to DNA fragments from different types of cancer cells.
The model correctly distinguished cancer patients at all stages from healthy individuals 91% of the time. Furthermore, the model can accurately identify stage 1 cancer patient samples in 87% of cases, suggesting that it may be able to detect early cancers.
Information-weighted fractionation of abnormal fragments has been shown to be suitable for detecting changes in tumor burden over time in confounding brain tumors such as glioblastoma, and this It could also provide real-time efficacy assessments of ongoing treatments for aggressive diseases,” said Michael Berens. He is a professor in the Brain Tumor Unit of the Institute for Genomics and a paper contributor.
Dr. Murtaza said the current results are encouraging, but more research is needed to improve the use of GALYFRE in patients of different age groups and with other medical conditions. The team also plans large-scale clinical studies to validate the test against certain cancer types, such as pancreatic cancer and breast cancer.
“One direction we are taking is to improve GALYFRE to make it more accurate for some patients who are at risk of developing certain types of cancer. Another aspect we are working on is determining whether our approach can be used to monitor treatment response in cancer patients undergoing chemotherapy.”
Murtaza said: “Further development will lead to a blood test for cancer detection and monitoring that will be clinically available in at least some conditions within the next 2-5 years and will eventually become available.” For patients with limited medical resources in the United States and around the world.”
