News release

From: Peter MacCallum Cancer Centre

AI as good as humans at predicting breast cancer outcomes

Two major international Peter Mac-led studies have found AI models were just as good as a human pathologist at a task which predicts outcomes for patients with breast cancer.

The studies conclude that AI models which count tumour-infiltrating lymphocytes (TILs) in samples of breast tissue should be more widely used “particularly where routine or widespread pathologist assessment is unavailable”.

TILs are immune cells that can be counted - usually by a pathologist looking at a microscope slide - and higher levels indicate a stronger immune response against the tumour. Previous studies have shown patients with more TILs in their breast tissue have better outcomes across several types of breast cancer.

Peter Mac’s Professor Sherene Loi led the two new studies and their findings – based on data from more than 5,600 patients with two major forms of breast cancer - published simultaneously this week in The Lancet Oncology.

The first study (CATALINA) compiled results from seven clinical trials involving patients with triple-negative breast cancer, and more than 1,300 of the patients had their TILs scored by both a pathologist and an AI model.

It found while AI and pathologist scores were not identical, both provided similar information about the patients’ prognosis. Patients whose tumours contained higher levels of TILs had better outcomes regardless of whether the immune cells were assessed by a pathologist or by AI.

The second study analysed more than 4,300 tumour samples from a phase III trial (APHINITY) involving patients with early HER2 positive breast cancer. TILs were assessed using expert pathologist scoring, digital image analysis and AI based methods. All three showed patients with more TILs had better outcomes, and the analysis went further.

Patients with the highest TILs - particularly those with node positive breast cancer – appeared to grain greater benefit from adding the drug pertuzumab to standard of care. AI could also analyse how immune cells were organised and highlight “hotspots” which provided additional prognostic information beyond simply counting TILs.

“Taken together, these two studies provide strong evidence that the immune response visible on a routine breast cancer pathology slide contains clinically important information. What is particularly reassuring is we see this across two very different types of breast cancer and across thousands of patients treated in randomised clinical trials,” Prof Loi says.

The studies also strengthen the case for TILs to be used as a practical biomarker in breast cancer, as well as showing that AI could enable their assessment at scale.

“Pathologist assessment of TILs remains highly reproducible when performed using standardised methods, but AI gives us the opportunity to measure this biomarker objectively and at enormous scale, including in settings where expert pathology assessment may not be readily available,” Prof Loi says.

“AI may also allow us to extract information that the human eye cannot readily quantify, such as the spatial organisation of immune cells within a tumour. Ultimately, combining pathologists and computational approaches may give us more information than either approach alone.”

The papers are “Artificial intelligence-based tumour infiltrating lymphocyte quantification in patients with triple-negative breast cancer: an independent validation study” (CATALINA) and “Manual, digital, and AI tumour-infiltrating lymphocyte scoring: a secondary analysis of the APHINITY randomised trial” (APHINITY).

The papers are also summarised in a podcast also now published on The Lancet Oncology website – Listen to the podcast “Tumour-infiltrating lymphocytes in breast cancer with Professor Sherene Loi”.

At over 20,000 cases a year, breast cancer is Australia’s most common cancer in women. About 15% are triple negative breast cancers, which are usually more aggressive and occur earlier, whilst 15-20% are HER2-positive.

< ends >

Contacts:

For more information contact the Peter Mac Communications team on 0417 123 048.

About Peter Mac

Peter MacCallum Cancer Centre is a world leading cancer research, education and treatment centre and Australia’s only public health service dedicated to caring for people affected by cancer.