Researchers from the University of Wisconsin–Madison have developed a technique that uses machine learning to identify specific types of cancer.
The approach involves analyzing short strands of DNA found in the blood of cancer patients. It’s a non-invasive method that can help doctors make a diagnosis and decide on the most suitable treatment.
A Closer Look at Liquid Biopsies
This breakthrough technique fits well with a process called “liquid biopsy”. Liquid biopsy uses simple blood draws instead of extracting cancerous tissue from a tumor with a needle.
According to Marina Sharifi, a professor and oncologist at UW–Madison, liquid biopsies are less invasive and can be conducted multiple times to track the cancer’s progress and response to treatment.
Cancerous tumors release a type of genetic material, known as cell-free DNA, into the bloodstream. But only specific parts of a cancer cell’s DNA are likely to break away.
The cells shield some of their DNA by bundling it up in secure balls called histones. These can be unwrapped to access sections of the genetic code when required.
The Role of Machine Learning
The research team, led by UW–Madison professors Shuang Zhao and Joshua Lang, used blood samples from nearly 500 patients, some with cancer and some without.
Kyle Helzer, a bioinformatics scientist at UW–Madison, explained that the sections of DNA containing genes that cancer cells use often are uncoiled more frequently and thus are more likely to break apart.
They split the samples into two groups. The first group was used to train a machine-learning algorithm to identify patterns among the fragments of cell-free DNA.
These patterns are like unique fingerprints specific to different types of cancer. The second group was used to test the trained algorithm.
The algorithm was able to identify both the presence of cancer and the specific type of cancer in a patient with more than 80% accuracy.
Implications for Cancer Treatment
This innovative technique was able to distinguish between two types of prostate cancer: the common adenocarcinoma and a fast-progressing variant called neuroendocrine prostate cancer (NEPC) that resists standard treatment methods.
Distinguishing NEPC from adenocarcinoma is often difficult, but the new approach helps doctors like Lang and Sharifi make the right diagnosis.
The machine learning method also offers advantages over traditional biopsy.
According to Jamie Sperger, another scientist in the study, liquid biopsies don’t require knowledge of the exact tumor location to be biopsied, and a standard blood draw is easier for the patient.
The study provides hope for better, personalized cancer treatment. It points towards the possibility of using existing gene sequencing technology to identify cancer types, potentially reducing time and cost.
This breakthrough could be a giant leap in diagnosing and treating cancer, taking us one step closer to personalized medicine.
If you care about cancer, please read studies that a low-carb diet could increase overall cancer risk, and berry that can prevent cancer, diabetes, and obesity.
For more information about health, please see recent studies about how drinking milk affects the risks of heart disease and cancer and results showing vitamin D supplements could strongly reduce cancer death.
The study was published in the Annals of Oncology.
Follow us on Twitter for more articles about this topic.
Copyright © 2023 Knowridge Science Report. All rights reserved.
