AI Pathology Tool Stratifies Rectal Cancer to Guide Chemoradiotherapy

Choosing intensified regimens for locally advanced rectal cancer is challenging because these therapies can cause serious side effects. Colorectal cancer is the fourth-most fatal cancer in the UK, and advanced stages recur frequently, prompting consideration of more aggressive treatment. Clinicians therefore need reliable, pre-treatment markers to identify who is likely to benefit. A new study shows that artificial intelligence can stratify biopsy samples by tumor cell density to guide use of irinotecan added to standard chemoradiotherapy.

University College London researchers developed an artificial intelligence (AI) approach to quantify tumor cell density in routine rectal cancer biopsy slides. To support clinical use, the team also created a free online platform, Octopath, where clinicians can upload slides for automated analysis. When applied to diagnostic slides, the method assigns patients to high- or low-density groups before treatment begins.

The AI identifies tumor regions within digitized histology and distinguishes cancerous from healthy cells. By counting millions of cells across whole-slide images, it generates a quantitative estimate of tumor cell density. This automated classification replaced a previously time-consuming manual process and enabled analysis at a scale suitable for large studies.

The evaluation formed part of a post-hoc analysis of the Phase 3 ARISTOTLE trial, which enrolled patients from 75 UK hospitals to test whether adding irinotecan to standard chemoradiotherapy improves outcomes. Investigators analyzed 414 pretreatment biopsy slides, classifying 188 as having high tumor cell density and 226 as having low tumor cell density.

Among patients with high-density tumors, adding irinotecan reduced the risk of recurrence by about 43% and the risk of death by about 50% compared with capecitabine-based chemoradiotherapy alone, while no benefit was seen in the low-density group.

Findings were published in EBioMedicine on July 27, 2026, as a post-hoc study of the Phase 3 ARISTOTLE trial. The authors noted that additional independent verification and clinical studies are needed before the AI can inform clinical decision-making or the intensified regimen is widely available to patients. The work was provided by University College London.

“Intensifying already taxing treatments puts additional strain on patients suffering from cancer. Clinicians need reliable ways to identify who is most likely to benefit before such treatments begin so potential side effects are avoided. Our findings show that doctors assisted by AI can pinpoint which patients will likely benefit from the more intensive treatment before it begins,” said Professor Maria Hawkins, senior author, UCL Medical Physics & Biomedical Engineering and clinician at UCLH.

Related Links
University College London