Colorectal cancer (CRC) is the one of the commonest causes of cancer worldwide, and, subsequently, there has been a drive in recent years to identify a non-invasive CRC biomarker. Volatile organic compounds (VOC) detection in various bodily substances, by means such as mass spectrometry and electronic nose, have gained particular interest. CRC patients have been shown to be distinguishable from healthy controls using urinary VOC detection in several studies, including two published by the research group at UHCW and the University of Warwick.1 2 There has also been significant research into the role that the intestinal microbiome plays in health and disease in humans.Introduction
The aim of this study was to characterise the urinary VOC and stool microbiome profiles of CRC patients, their spouses and first degree relatives with the goal of determining whether environmental and genetic controls could be distinguished from the CRC subjects using urinary VOC and faecal microbiome profiling.Methods
56 CRC subjects, 45 spouses and 37 relatives were recruited. Sample analysis was performed using an LC-FAIMS-MS apparatus to detect urinary VOCs, whilst an Illumina Miseq platform was used for 16 s RNA sequencing. Urinary data was processed and analysed using a 5-fold cross-validation with sparse logistics regression and random Forrest statistical classifiers. Microbiome data was analysed using standardised uPARSE and QIIME pathways. Comparisons were also made between pre-treatment and post-treatment CRC samples (n=23) to determine if there was any change in VOC or microbiome profiles after treatment.Results
The urinary VOC profiles of CRC subjects could be distinguished from both sets of healthy controls using both classifiers. Achieved sensitivities were 63%–69%, specificities 64%–69% and AUC 0.71–0.72. No statistically significant differences could be found in the urinary VOC profiles of pre-operative and post-operative samples.Results
Microbiome analysis revealed over 1300 operational taxonomic units (OTUs), with a similarity of >93% between CRC samples and control groups, with significantly different bacterial abundances identified in 82 OTUs, mainly Clostridiales. Pre-treatment and post-treatment sample analysis revealed differences of 17 (3%) and 22 (4%) OTUs at 3 and 6 months respectively, again principally Clostridiales.Conclusions
This study provides further validity of the use of urinary VOCs as a non-invasive biomarker for CRC detection, demonstrated here against genetic and environmental controls. The LC-FAIMS-MS technology is a variant of the previously utilised FAIMS, although exact chemical identification is not possible due to a lack of a validated database. Microbiome analysis showed broadly similar bacterial profiles between the various groups, will subtle differences in some families, such as clostridiales, and a restricted CRC profile, compared to the healthy controls.