Stay in touch
We'd love to keep in touch about news, events and how you can get involved. To hear from us, please sign up below.
Dr Kastytis Sidlauskas, Prof Louise Jones and Prof Nasir Rajpoot are using artificial intelligence to develop a better way to know how likely DCIS is to become invasive breast cancer. It could mean that some people can safely avoid treatment and its side effects.
Researcher: Dr Kastytis Sidlauskas
Location: Barts Cancer Institute, Queen Mary University of London
Project cost: £346,116
Ductal carcinoma in situ, or DCIS, is an early form of breast cancer. In DCIS, all breast cancer cells stay inside a breast duct because they don’t have the ability to spread into the surrounding breast tissue. Everyone diagnosed with DCIS receives treatment. This is to stop DCIS becoming invasive breast cancer. But we know that around half of all DCIS cases will not become invasive and they may not cause any harm. We currently don’t know how to tell apart DCIS cases that will become invasive breast cancer and those that won’t. Knowing which DCIS cases aren’t harmful could help people skip treatment which may be unnecessary and spare them its side effects.
‘Our aim is to reduce treatment for patients with DCIS at low risk of becoming invasive cancer. Currently, all women with DCIS receive surgical treatment, and sometimes also radiotherapy and hormone therapy. Clinical trials are underway in which patients with low or intermediate grade DCIS can choose to be monitored instead. But basing these decisions on grade alone may not be that accurate. We need to find the risk features that would help to tell more accurately if DCIS is likely to cause harm,’ Dr Kastytis Sidlauskas
The science behind the project
DCIS grade alone, as it is currently assessed, may not be the most accurate way to determine who could safely skip treatment. In this project, Dr Kastytis Sidlauskas supervised by Prof Louise Jones and Prof Nasir Rajpoot are using artificial intelligence (AI) to more accurately identify DCIS grade and to capture other features in DCIS that can determine its aggressiveness.
Kastytis is analysing breast tissue images containing DCIS and invasive breast cancer cells. For the analysis, he is using sophisticated AI tools developed by the leading experts in computer science. These tools can detect the features of breast cells that aren’t detectable during standard analysis. He will then work out which of these features are linked to DCIS becoming invasive breast cancer. For this research, Kastytis is using DCIS samples from several clinical trials.
He will combine these findings with existing knowledge of other features that affect the behaviour of DCIS. It will help Kastytis develop a more accurate method to assess how likely DCIS is to become invasive breast cancer. In the future, this approach could be used in clinics to determine who needs immediate treatment and who could safely skip it and be monitored instead.
This research has the potential to change how DCIS is treated. It could allow some people to safely skip treatment and its associated side effects, while making sure those who need it receive treatment and their DCIS doesn’t come back as invasive breast cancer.
Make a donation to support our research