RCSI research could spare breast cancer patients from unnecessary chemotherapy

The problem: too many people getting chemo 'just in case'

Chemotherapy is routinely used to treat early-stage, hormone receptor-positive, HER2-negative breast cancer, which makes up around 70% of all breast cancer diagnoses each year. It remains an important treatment, but its side effects can be debilitating, and for a significant number of patients doctors simply can't be sure whether the benefit outweighs the harm.

The issue is that current genomic risk scores often come back 'intermediate', which leaves both doctor and patient in a grey zone. In practice, that usually means chemotherapy gets prescribed as a precaution, even though some of those patients would have stayed cancer-free without it. The genomic test most widely used for this, the Oncotype DX Recurrence Score, is helpful precisely because it works well at the extremes, flagging patients at clearly low or clearly high risk, but the study authors note that it is far less useful for the sizeable group of patients who fall into the score's middle band, roughly 16 to 25 on the scale.

Patients in that intermediate group are also the ones most likely to experience a late relapse, sometimes five to ten years after their original diagnosis, which is part of what makes the treatment decision so fraught: doctors are essentially being asked to weigh up a risk that may not show up for years against side effects that start immediately. That mismatch between an early decision and a long-term outcome is exactly the gap this new research is aimed at closing.

What RCSI and UCD found

Researchers from RCSI University of Medicine and Health Sciences and University College Dublin, publishing in Nature Communications in June 2026, found that analysing a patient's own immune cells near the tumour using new AI-based methods can add valuable extra insight to identify who can safely skip chemotherapy.

The team used samples from a randomised Irish trial comparing hormone-blocking therapy alone against hormone-blocking therapy combined with chemotherapy, specifically in patients with intermediate genomic risk scores. They found that the density of cancer-fighting cytotoxic T-cells around a tumour was a strong predictor of how a patient would respond to treatment, with patients who had high levels of these T-cells actually showing poorer outcomes from chemotherapy, a striking and clinically useful signal.

The study, run across 440 tissue samples using spatial proteomics and a further 359 samples using spatial transcriptomics, two techniques that map where specific molecules and gene activity sit within a tumour rather than just measuring them in bulk, found that this pattern held up when the researchers checked it against a separate, independent set of whole-tumour samples. That kind of validation in a second dataset is exactly what gives a biomarker finding like this real weight, rather than leaving it as an interesting result from a single study.

Professor Darran O'Connor, research lead at RCSI's School of Pharmacy and Biomolecular Sciences, said that for patients with an intermediate genomic risk, the decision around chemotherapy is often difficult, and uncertainty frequently leads to treatment that may not have been necessary, affecting quality of life. He added that genomic testing has already advanced doctors' ability to tailor treatment for these patients, but that AI-based analysis of the tumour microenvironment can now add a further, more precise layer of information on top of that.

The road from lab bench to clinic

Professor William Gallagher of UCD's School of Biomolecular and Biomedical Science and UCD Conway Institute, senior author on the study, said the AI-based method represents an important step towards more personalised cancer care, but was clear that further validation in larger studies will be needed before it can be used in everyday clinical practice. He said the findings give researchers a clearer picture of what drives recurrence risk in patients with intermediate genomic scores. Dr Zak Kinsella, first author on the study and a postdoctoral researcher at RCSI, said the findings bring the field closer to the kind of personalised treatment decisions that could avoid unnecessary chemotherapy.

This work sits alongside a broader push in Irish breast cancer research. Through the Precision Oncology Research Initiative for Metastatic Breast Cancer, led by the Beaumont RCSI Cancer Centre, researchers separately identified a gene regulator called CDK12 as a promising new drug target for advanced, hormone receptor-positive breast cancers that have become resistant to treatment, work published in the Journal of the National Cancer Institute and now being taken forward with industry partner Carrick Therapeutics. Together, these strands point toward a future where treatment for breast cancer, including its hardest-to-treat forms, is increasingly tailored to the biology of each patient's tumour rather than a one-size-fits-all protocol.

RCSI's cancer research isn't limited to breast cancer either. In September 2026, four RCSI researchers were awarded more than €1.2 million through the Health Research Charities Ireland-Health Research Board Joint Funding Scheme, including funding for a project using a simple stool test to identify colorectal cancer earlier in younger adults, and a project investigating how cancers spread to the brain with a view to earlier detection. That signals a steady pipeline of Irish cancer research still coming through behind this latest advance.

The multidisciplinary nature of the breast cancer study is also worth noting: the research team drew on immunology, computational biology and clinical oncology expertise across both RCSI and UCD, and made use of samples originally gathered for a randomised clinical trial rather than a freshly recruited cohort, meaning the underlying patient data had already been through the rigour of trial-level collection and follow-up. That kind of collaboration, building new AI-driven insight on top of existing, well-characterised trial data, is increasingly how Irish cancer researchers are trying to get more clinical value out of studies that have already been run, rather than always needing to start from scratch.

The RCSI and UCD findings also sit within a fast-moving international field working on the same problem from a different angle. Research presented at the American Society of Clinical Oncology's 2026 annual meeting used deep learning applied to routine tumour tissue slides to predict chemotherapy benefit specifically in premenopausal women with intermediate genomic scores of 16 to 25, the same difficult-to-call group at the centre of the Irish study. Built on data from the original US TAILORx trial, that model was able to sort patients into distinct low-risk and high-risk groups with different outcomes, echoing the central idea behind the Irish team's work: that squeezing more information out of tumour samples doctors already have, rather than waiting for a new test, can sharpen decisions that have long rested on a single genomic score.

Dr O'Connor has also been candid about what the new AI method does and doesn't replace. Speaking to the Irish Examiner, he described the approach as one that could, in principle, be run in any hospital with 'a decent laptop', since it builds on an adapted open-source tool rather than proprietary equipment, but stressed that a pathologist still needs to be 'in the loop' interpreting the results rather than leaving the decision to software alone. That framing, of AI as a second opinion that sharpens a specialist's judgement rather than replacing it, reflects how much of Ireland's cancer research community is now trying to introduce these tools into everyday practice.

The Bottom Line

This is exactly the kind of research that changes lives quietly rather than dramatically: not a new drug, but a better way of deciding who truly needs one. If larger studies confirm what this RCSI and UCD team found, it could mean fewer women going through the physical and emotional toll of chemotherapy they never needed, while making sure those who do benefit still get it. With further funding now flowing into related Irish cancer research, the pipeline behind this finding looks well placed to keep delivering.