AI helps identify new biomarkers for breast cancer

US scientists, led by Case Western Reserve University, have used artificial intelligence (AI) to identify new biomarkers for breast cancer that can predict whether the cancer will return after treatment — which can be identified from routinely acquired tissue biopsy samples of early-stage breast cancer. The key to this initial determination is collagen — a common protein found throughout the body, including in breast tissue.

Previous research had suggested that the collagen network, or arrangement of the fibres, relates strongly to breast cancer aggressiveness. Researchers have now used standard tissue biopsy slides and AI to definitively demonstrate the critical role played by collagen, with their results published in the journal npj Breast Cancer.

The researchers, using machine-learning technology to analyse a dataset of digitised tissue samples from breast cancer patients, were able to prove that a well-ordered arrangement of collagen is a key prognostic biomarker for an aggressive tumour and a likely recurrence. Conversely, they showed that a disordered or broken-down collagen infrastructure not only indicates a better outcome, it actually promotes one. They also found that the disordered collagen network prevents an otherwise aggressive tumour from migrating out of the breast tissue and helps prevent its return after various cancer treatments like chemotherapy.

“It sounds counterintuitive, but the collagen fibres play a role in tumour migration,” said Professor Anant Madabhushi, Head of Case Western’s Center for Computational Imaging and Personalized Diagnostics (CCIPD). “One way to understand it is to say that if the collagen ‘highway’ is in terrible shape, it’s more difficult for the tumour to migrate, but if it’s smooth and organised, it makes it easier for the tumour to hitch a ride.”

Doctoral student Haojia Li, who led the research, said the project was important because it validates previous findings that suggested highly organised collagen indicates a worse prognosis. Furthermore, it was accomplished with digitised images of simple tissue slides, suggesting this method could become part of a pathologist’s routine. Current methods for examining and investigating the collagen architecture require an expensive and less common electron microscope.

“Our method would make predicting outcomes much more available to more doctors and in hospitals which don’t have the resources to have an advanced imaging microscope,” Li said. “That’s why this is so exciting — because it can give the physician the information he or she needs to guide how aggressively to treat the cancer.”

Prof Madabhushi said that because the models built by his team were validated on a completed clinical trial dataset, it would “provide a higher level of evidence with regard to the validity of the collagen signature” and function as a “natural segue into prospective clinical trial validation”.

Image credit: ©stock.adobe.com/au/Axel Kock

Please follow us and share on Twitter and Facebook. You can also subscribe for FREE to our weekly newsletters and bimonthly magazine.