A New Approach to Understanding Rectal Cancer in the United Kingdom

A research group in the United Kingdom has developed an artificial intelligence tool capable of analysing routine pathology slides from rectal cancer patients to identify immune patterns linked to treatment response and risk of recurrence. The findings, published in eBioMedicine, highlight a potential shift in how clinicians assess disease progression and personalise treatment strategies.

Rectal cancer outcomes are strongly influenced by the immune environment surrounding the tumour. However, conventional methods for studying these interactions—such as genomic sequencing or advanced spatial imaging—are costly, complex, and time-intensive. The team from UCL Medical Physics & Biomedical Engineering, therefore, explored whether AI could extract the same level of biological detail directly from the microscopy images already produced during standard diagnosis.

AI Detects Immune Cell Signatures in Minutes

Traditionally, pathologists examine tumour slides manually, but the researchers demonstrated that AI can quickly recognise patterns involving key immune cells such as lymphocytes and macrophages. According to senior author Dr Charles-Antoine Collins-Fekete, AI is capable of extracting clinically useful information “in minutes”, offering a faster and more affordable alternative to high-cost molecular testing.

The research team trained the system on millions of pathology images and later validated it using samples from three distinct patient cohorts, including individuals enrolled in the ARISTOTLE clinical trial. Approximately 900 patient samples were included in the evaluation.

Immune Activity Predicts Survival and Recurrence

The analysis confirmed long-standing observations in cancer immunology:

  • Higher levels of lymphocytes in the tumour microenvironment were associated with better survival and reduced risk of recurrence.

  • Increased macrophage presence, particularly certain subtypes known to support tumour growth, correlated with poorer outcomes.

These immune characteristics are not currently used as standard decision-making tools in rectal cancer management in the UK, but the findings suggest they could help determine which patients will benefit most from chemoradiotherapy.

The AI tool could also assess immune changes after treatment. Patients whose tumours showed an increase in tumour-infiltrating lymphocytes after therapy generally experienced better outcomes, consistent with the known ability of chemoradiotherapy to stimulate local anti-tumour immunity. Conversely, tumours that remained immunologically inactive were associated with earlier recurrence.

Genetic Mutations and Immune Response

The study also examined how genetic factors influence immune activity. Patients with:

  • A normal KRAS gene and high lymphocyte counts demonstrated superior survival rates.

  • TP53 mutations combined with elevated macrophage levels tended to have significantly poorer outcomes.

These findings echo previous international research showing that both KRAS and TP53 mutations can alter the immune response and influence treatment sensitivity.

Cell Growth Rate as an Additional Prognostic Marker

Tumours exhibiting high mitotic activity—an indicator of rapid cell division—were found to be less responsive to immune surveillance. High proliferation rates were strongly associated with more aggressive disease and earlier recurrence, reinforcing the importance of understanding tumour biology beyond traditional staging.

A Free Clinical Tool for Health Professionals

To support translation of their work into clinical practice, the researchers developed Octopath, a free online platform that allows clinicians to upload pathology slides and receive automated immune analyses. While still in early development, the tool aims to help oncologists better understand tumour behaviour and refine treatment plans.

However, the team emphasised that further research involving larger and more diverse patient groups is necessary before widespread clinical adoption.

Looking Ahead: AI’s Role in Cancer Classification

Senior author Professor Maria Hawkins, consultant clinical oncologist at UCLH, described the study as an important step toward integrating AI into cancer classification. She noted that while the technology is not yet ready to guide treatment independently, its potential to support clinical decision-making is both “promising and exciting”.

The researchers plan to explore additional immune cell types and apply more sophisticated analytical techniques to deepen understanding of how cancer interacts with the immune system—a field likely to influence future personalised oncology across the United Kingdom and beyond.