A study to assess the clinical and cost-effectiveness of the Ibex Artificial Intelligence Histology system in diagnosing clinically important prostate cancer on prostate biopsy tissue

Background About 80,000-100,000 men every year undergo prostate biopsy to diagnose prostate cancer, equating to 4 million histology slides. This is predicted to double by 2030.

Combined with a pathology work-force shortage, and variability in pathology reporting, an artificial intelligence (AI) tool might be beneficial to the NHS. The Ibex-AI histology system has shown high sensitivity and specificity in other healthcare settings.

Aims To determine the clinical and cost-effectiveness of the Ibex-AI system in ruling-out and detecting clinically important prostate cancer on scanned images of pathology slides processed in NHS hospitals.

Plan The pathology slides from 600 men will be prospectively scanned at high resolution for the Ibex-AI system to analyse in parallel to the pathology review. Both will be blinded to the findings of the other. Discordant slides will be reviewed by 3 independent expert uro-pathologists with immunocytochemistry.

We will assess sensitivity, specificity, negative and positive predictive values of Ibex-AI in detecting and ruling-out any Gleason 3+4=7 prostate cancer as well as any cancer and other thresholds of clinical importance based on grade and length of cancer. Timelines for Delivery Our proposal is for 2-years.

In the 3 months prior to starting, we will draft trial documents. Regulatory approvals and site set-up will occur in the first 3-4 months and recruitment start by month 5. The calibration study will be conducted over 4 months. The validation study will recruit from 6 sites over 14 months, leaving 2 months for data analysis and final report.

Anticipated Impact and Dissemination We will present our findings at conferences and publish in high impact journals.

We will share results with trial participants and key stakeholders like patient support groups, charities, policy makers and professional medical groups. Our findings will inform clinical guidelines.