Colorectal cancer (CRC) is the 2nd most common cause of UK cancer death and makes up 11% of new cancer cases. However, 5-year survival rates are 58%, amongst the lowest in Europe. There is need for improvement because, if detected early, CRC survival rates can exceed 90%. A test measuring breath volatile organic compounds (VOCs) could provide an attractive non-invasive tool for early CRC detection, aiming to augment and streamline existing referral pathways, and ultimately save lives.
The first aim of this PhD was methodological development aiming to improve reliability and standardisation of measurement and interpretation of breath VOCs. This culminated in the development of a quality control system, providing a framework for detection of inadequate quality breath samples prior to analysis. The second aim was to determine the diagnostic accuracy of breath for CRC and adenomatous polyp diagnosis, using the ReCIVA® breath sampler and thermal desorption tubes. 1432 patients having a colonoscopy or colorectal adenocarcinoma resection over 7 London hospitals had their breath analysed using gas chromatography mass spectrometry (GC-MS). Machine learning algorithms showed that CRC patients (n=162) could be predicted from positive and negative controls (n=1270) with a sensitivity of 77%, specificity of 87%, and an area under the Receiver Operating Characteristic (AUROC) curve of 0.87. Fifteen of the most predictive VOCs were from sulphur containing, alcohol, alkane, ester or phenol chemical groups. Colorectal polyp patients (n=592) could be predicted from positive and negative controls (n=619): sensitivity 66%, specificity of 54%, AUROC curve of 0.65. The third aim was to demonstrate feasibility and acceptability of breath testing in a large primary care study.
These exciting findings represent advancement in knowledge of diagnostic CRC VOCs, paving the way for larger targeted studies to validate findings, establish quantitative thresholds and move towards eventual implementation of breath testing into clinical practice.