COLOSPECT

Chief Investigator: Greg Taylor

Summary

Phase two prospective observational single blind multi-site performance evaluation trial of serum Raman spectroscopy and supervised machine learning classification in a FIT-positive bowel screening population

Funder: Cancer Research Wales

Background

Bowel screening seeks to detect early stage bowel cancers and the bowel polyps that can lead to cancers. Removal of these higher risk adenomatous polyps is considered to prevent cancer development. This is done through a colonoscopy procedure which requires bowel preparation and time off work. Rather than using colonoscopy itself as the screening tool, the UK bowel screening committees recommend a pre-screening test to determine the need for colonoscopy. This is a faecal based home test that detects human haemoglobin that can arise from cancers and some polyps. If the faecal immunochemical test (FIT) is above a certain threshold then a colonoscopy will follow. Studies show that the pooled sensitivity of FIT was 79% (95% confidence interval 0.69–0.86) for CRC and specificity 94% (95% CI, 0.92–0.95). FIT therefore has a known false negative rate (misses cancers and polyps) and false positive rate (FIT is positive but the colonoscopy is normal). FIT is known to miss at least 50% of high risk adenomas. After a positive FIT, even at high detection thresholds, up to 30% of subsequent colonoscopies are normal and are potentially unnecessary.
Raman spectroscopy analysis of blood serum has shown good levels of accuracy for detecting colorectal cancer and adenomas in earlier work. The aim of this study is to assess the test performance of a Raman spectroscopy analysis routine in a FIT positive screening population.

Method

This study will recruit screening participants having colonoscopy from the screening assessment centres across Wales. It seeks to obtain 2,000 samples for analysis using spectroscopy in vitro diagnostic devices. Outcomes from colonoscopy (cancer/ adenoma/ normal findings) will be used to train and test a machine learning algorithm to allow calculation of the sensitivity and specificity of the device.

Impact

The addition of a cancer specific test such as Raman to the non-cancer specific FIT test would increase the pre-test probability of cancer/adenoma, which could reduce the missed cancer rate at time of colonoscopy (current post-colonoscopy colorectal cancer (PCCRC) rates are 8.6% in the English NHS), and assist with prioritization of screening colonoscopies at times of high demand.


Contact Details

Trial Manager / Lead Contact: Dr Nicola Heady
Phone: 01792 604905
Email: n.j.heady@swansea.ac.uk

IRAS ID: 293364;