Could data predict if breast cancer will return?

Every year, thousands of New Zealanders have their breast cancer details added to Te Rēhita Mate Ūtaetae – Breast Cancer Foundation National Register. The Register is a taonga, and holds data gifted by over 60,000 breast cancer patients across Aotearoa, past and present. This data is held securely and is anonymised for use in research, to audit care, to help plan breast cancer services, and to identify where further progress is needed to improve the lives of everyone affected by breast cancer.

When a breast cancer diagnosis can feel overwhelming and uncertain, being part of the Register may be far from people’s minds. But the information people gift us during this difficult chapter has a lasting impact that reaches far into the future, helping people diagnosed with breast cancer years from now.

For pathologist Dr Gavin Harris, the Register is the reason ground-breaking research can occur, right here in New Zealand.

Gavin is a UK-trained anatomical pathologist who has worked in New Zealand for more than 20 years. He moved here in 2003 to develop the breast pathology service at Christchurch Public Hospital and helped establish the Christchurch Breast Cancer Register in 2009, as he had seen international examples of what good data could achieve. Data in the Register has been collected since 2000, and Gavin now sits on the Register’s Clinical Advisory Group.

"You can't make improvements unless you know where the issues are," he says. “And because New Zealand has unique demographics, you can’t base your research on overseas data. That’s what makes the Register an unparalleled resource, as the data is representative of our population, including our Māori and Pasifika communities”.

Right now, Gavin is leading research into something that could change the lives of people with breast cancer: teaching computer algorithms to read breast cancer tissue slides using machine learning/artificial intelligence approaches. The aim is to be able to predict whether a person's cancer is likely to come back, years down the track. At the moment the focus is on the subtype that makes up approximately 70% of invasive breast cancers (oestrogen receptor-positive/HER2-negative).

This matters because when breast cancer returns, it's often because it has spread elsewhere in the body (metastasis), making it harder to treat. It’s difficult to know at the time of diagnosis who is at risk of their cancer coming back. Some people may end up receiving more treatment than they need, while others may need closer monitoring or additional therapy. If Gavin's algorithm can reliably predict this, it could change how breast cancer is treated across New Zealand - helping doctors match the treatment to each person's real risk and help more people live well beyond a breast cancer diagnosis. It's the kind of practice-changing research that brings us closer to a future where no one dies from breast cancer.

Currently breast cancer tissue is placed on a glass slide, stained with a special dye and read under a microscope. In some situations, the physical glass slides need to be sent overseas for special analysis, which costs of thousands of dollars, and means waiting weeks for an answer. Gavin's team of researchers are looking at getting that same analysis done by scanning those slides into digital images and analysed using computer algorithms, in a fraction of the time, at minimal cost.

But to develop this technology, the researchers needed something only the Register could provide.

To teach the computer algorithms what to look for, Gavin's team needed real proof of what happened to real people with breast cancer over many years, not a guess. That's exactly what the Register provided: detailed, anonymised, long-term records showing whose cancer came back, and when.

"This sort of research couldn't happen without the Register," Gavin says. "The clinical data from the Register is what provides ground truth for the algorithm's training. That becomes the truth, and the algorithm has to adjust itself to pick up the features that predict whether the cancer will come back."

The team is finishing its analysis now, with early results expected by the early 2027. The next step is testing the approach using a different set of samples, scanned on different scanners in different laboratories to compare the results.

The benefits could be significant. Not only could patients avoid the cost and delay of sending samples overseas, but results could be measured in hours rather than weeks. Once the results are ready, multidisciplinary team meetings where surgeons, oncologists and pathologists meet together, can decide on the treatment pathway for the patient based on their results.

Your support makes this possible. Every donation that helps fund the Register is helping build a resource researchers like Gavin rely on, one that's already opening doors nobody thought possible for New Zealand.

"It really is a unique, unparalleled resource. Being able to access this data so quickly is absolutely phenomenal. It's all been developed by the Foundation with advice from clinicians - I can't speak highly enough of it," Gavin says. "My focus has always been on improving outcomes for patients, and you can't do that without something like the Register."