Search in the Abstract Database

Abstracts Search 2019

P842 Multi-omics analysis suggests an active role of fungi in Crohn’s disease

A. Frau*1, U. Z. Ijaz2, R. Hough1, B. J. Campbell1, J. G. Kenny3, N. Hall4, J. Anson5, A. C. Darby3, C. S. J. Probert1

1University of Liverpool, Cellular and Molecular Physiology, Liverpool, UK, 2University of Glasgow, School of Engineering, Glasgow, UK, 3University of Liverpool, Centre for Genomic Research (CGR), Liverpool, UK, 4Earlham Institute, Norwich, UK, 5Royal Liverpool and Broadgreen University Hospitals NHS Trust, Liverpool, UK


Several studies have suggested a role of fungi in Crohn’s disease (CD); from the reporting of ASCA in CD patients, the observation of fungal metabolites in patients in relapse, to recent mycobiome studies. However, the analysis of the gut mycobiome is made difficult by a very low diversity, combined with a relevant input of fungi from the diet. Therefore, discriminating between active and transient fungi using only metagenomics is not easy. To overcome this issue, we combined metabolomics data, which also indicate microbial activity, with bacterial and fungal communities’ data.


Briefly, we produced volatile organic compound metabolomic data along with bacterial 16S rRNA and fungal 18S rRNA data from 43 donors (23 CD patients and 20 controls). These data were filtered and normalised and DIABLO (MixOmics), a statistical tool which integrates omics data, was used. This uses supervised analysis to highlight signature features and to identify correlated variables.


We compared CD patients vs. controls and CD active patients vs. controls. The first comparison gave a balanced error rate (BER) of the cross-validation of the model < 35%. The model was made of two components, the first showed a higher Pearson correlation between Bacteria and VOCs (0.54). Meanwhile, the second showed a higher correlation between Fungi and Bacteria (0.66). We also found that branched-short chain fatty acids (high in CD) were correlated with bacteria OTUs assigned to gut fermenters, mainly Firmicutes. This result alone shows the potential of this approach to pinpoint microorganisms that are active in the gut of CD patients. The second comparison saw CD active (n = 11) vs. controls (n = 20). The difference in the number of samples gave a BER relatively high (around 45%). Again, two components were selected, but these gave a Pearson correlation between the omics higher than the previous comparison (up to 0.67 VOCs and Bacteria, 0.62 Bacteria and Fungi and 0.55 VOCs and Fungi). Correlation of variables showed that several OTUs assigned to Saccharomycetes yeasts and a mould (Aspergillus) were correlated to metabolites associated to fungi (e.g. heptanal and 3,7-dimethylocta-1,6-dien-3-ol), supporting a possible role of fungi in active CD. These fungi were also correlated to Clostridiales and Enterobacteriales.


The high BER do not allow us to draw definite conclusions and further studies, with a higher number of patients, are required. However, we can say that fungi are very likely to be active during relapse. We also show a powerful approach that allows to overcome the issues related to the interpretation of gut mycobiome studies, which are biased by the large input of yeasts from the diet.