Rectal cancer AI treatment analysis flags who may benefit from irinotecan

A rectal cancer AI treatment analysis by University College London researchers has identified a subgroup of patients who appeared to live longer when irinotecan was added to standard chemoradiotherapy. The finding matters because the intensified regimen can bring serious side effects, and doctors need stronger evidence before using it more widely.

The study, published in eBioMedicine, re-examined biopsy samples from the phase III ARISTOTLE trial, which enrolled patients at 75 UK hospitals. That trial tested whether irinotecan could improve outcomes for people with locally advanced rectal cancer, meaning disease that has started to spread into nearby tissue.

According to UCL, the original trial found little overall benefit from adding irinotecan. The new analysis used artificial intelligence to separate patients by tumor cell density, revealing that the treatment effect differed by how many cancer cells were present in diagnostic biopsy samples.

How could AI guide rectal cancer treatment?

The AI system was trained to read microscopic biopsy images, distinguish tumor from surrounding tissue and identify cancerous and healthy cells, according to the researchers. By counting large numbers of cells in standard biopsy images, it classified patients into high- and low-density tumor groups before treatment began.

UCL said patients with high tumor cell density who received irinotecan in addition to standard chemoradiotherapy had about a 43% lower risk of cancer returning and about a 50% lower risk of death than those given the usual regimen of capecitabine chemotherapy plus radiation therapy. Patients whose biopsies showed low cancer cell density did not show a difference in outcomes.

The researchers applied the AI method to 414 rectal cancer biopsy slides from ARISTOTLE participants. The system placed 188 patients in the high tumor cell density group and 226 in the low-density group, according to the study.

UCL said manual measurement of tumor cell concentration had been too slow and impractical for a study of that size. The AI approach allowed the team to process biopsy images, count millions of cells and sort patients faster than conventional review, which helped uncover the treatment pattern.

Why irinotecan has been hard to use more broadly

Irinotecan is a chemotherapy drug already used in advanced bowel cancer, according to UCL. It has long been considered a possible addition to standard treatment for advanced rectal cancer, but earlier research had not shown improved patient outcomes across the full trial population.

Adding irinotecan to capecitabine may increase cancer-fighting activity, UCL said, but it can also intensify side effects such as diarrhea and low white blood cell counts. Because rectal cancer treatment is already demanding, the researchers said clinicians need reliable ways to identify patients likely to benefit before increasing treatment intensity.

Dr. Zhuoyan Shen of UCL Medical Physics and Biomedical Engineering, the study’s lead author, said the AI analysis distinguished patients who benefited from irinotecan from those who did not, showing that AI can reveal tumor biology that is hard to measure consistently with standard methods.

Professor Maria Hawkins of UCL Medical Physics and Biomedical Engineering and UCLH, the senior author, said the findings suggest AI-assisted doctors may be able to identify patients likely to benefit from a more intensive regimen before treatment starts.

The research team has made a free online tool called Octopath, where clinicians can upload biopsy slides for analysis. UCL cautioned that the findings still need independent verification and further clinical studies before the AI system can be used to make treatment decisions or the irinotecan combination becomes widely available for these patients.

This story draws on original reporting from Medical Xpress.