Downloaded from http://gut.bmj.com/ on December 14, 2017 - Published by group.bmj.com Irritable bowel syndrome ORIGINAL ARTICLE An irritable bowel syndrome subtype defined by species-specific alterations in faecal microbiota Ian B Jeffery,1 Paul W O’Toole,1,2 Lena O¨hman,3 Marcus J Claesson,1,2 Jennifer Deane,1,2 Eamonn M M Quigley,2 Magnus Simre´n3 < Additional materials are ABSTRACT published online only. To view Background and aims Irritable bowel syndrome (IBS) is Significance of this study these files please visit the a common functional gastrointestinal disorder that may journal online (http://gut.bmj. com/content/61/7.toc). be triggered by enteric pathogens and has also been What is already known on this subject? linked to alterations in the microbiota and the host < 1Department of Microbiology, Irritable bowel syndrome (IBS) onset may follow University College Cork, Cork, immune response. The authors performed a detailed enteric infections. Ireland analysis of the faecal microbiota in IBS and control < A low-grade inflammatory response has been 2 Alimentary Pharmabiotic subjects and correlated the findings with key clinical and described in IBS. Centre, University College Cork, physiological parameters. < Various changes in the microbiota have been Cork, Ireland Design The authors used pyrosequencing to determine 3Department of Internal described in IBS but their primacy has not been Medicine, Institute of Medicine, faecal microbiota composition in 37 IBS patients (mean defined. Sahlgrenska Academy, age 37 years; 26 female subjects; 15 diarrhoea- University of Gothenburg, predominant IBS, 10 constipation-predominant IBS and What are the new findings? Gothenburg, Sweden 12 alternating-type IBS) and 20 age- and gender- < A detailed assessment of the faecal microbiota in IBS does not reveal a uniform change in the Correspondence to matched controls. Gastrointestinal and psychological Dr Paul W O’Toole, Department symptom severity and quality of life were evaluated with microbiota. of Microbiology, University validated questionnaires and colonic transit time and < Analysis of microbial populations in IBS reveals College Cork, 447 Food Science rectal sensitivity were measured. distinct clusters, of which some overlap with Building, Cork, Ireland; Results Associations detected between microbiota normal controls while others are quite different. [email protected] composition and clinical or physiological phenotypes < An increased Firmicutes:Bacteroidetes ratio best Accepted 22 October 2011 included microbial signatures associated with colonic characterises those IBS subjects who differ Published Online First transit and levels of clinically significant depression in the from normal populations. 16 December 2011 disease. Clustering by microbiota composition revealed < The microbial signature is related to the clinical subgroups of IBS patients, one of which (n¼15) showed phenotype in a subset of IBS patients. normal-like microbiota composition compared with healthy controls. The other IBS samples (n¼22) were How might it impact on clinical practice in the defined by large microbiota-wide changes characterised foreseeable future? < by an increase of Firmicutes-associated taxa and Microbial fingerprinting may help to identify IBS a depletion of Bacteroidetes-related taxa. subpopulations with differing prognosis and Conclusions Detailed microbiota analysis of a well- varied therapeutic responses. characterised cohort of IBS patients identified several clear associations with clinical data and a distinct subset of IBS patients with alterations in their microbiota that infectiondpostinfectious IBS. Postinfectious IBS did not correspond to IBS subtypes, as defined by the has been associated with a number of bacterial e Rome II criteria. species.2 4 This and the hypothesis that small intestinal bacterial overgrowth might also contribute to IBS symptomatology5 have moti- vated investigation of antibiotics as therapeutic INTRODUCTION agents in IBS.6 Large double-blind studies of some Irritable bowel syndrome (IBS) is a common antibiotics have shown an improvement (albeit gastrointestinal (GI) disorder that affects 10e20% modest, though statistically significant) in IBS of adults in industrialised countries.1 Although it symptoms. The improvement has been attributed has a non-fatal prognosis, IBS patients report more to a reduction in total bacterial load and, perhaps, ailments than the general population and an associated suppression of certain causative commonly incur a diminution in quality of life.1 bacterial species.67 Characteristic symptoms include abdominal pain The gut microbiota of IBS subjects has thus and/or discomfort, bloating, distension and altered attracted considerable interest, and alterations in bowel habits.1 microbiota composition have been linked to IBS No single unifying cause has been identified for and IBS-related symptoms. To cope with the IBS, but there is recent evidence suggesting the heterogeneity in IBS symptoms,8 these studies involvement of the gut microbiota. The clearest often employ the Rome criteria to identify evidence is provided by subjects who develop IBS subtypes (reviewed by Salonen et al).9 Their IBS de novo following an episode of enteric meta-analysis of IBS microbiota, as determined by Gut 2012;61:997e1006. doi:10.1136/gutjnl-2011-301501 997 Downloaded from http://gut.bmj.com/ on December 14, 2017 - Published by group.bmj.com Irritable bowel syndrome high-throughput methods, identified consistent microbiota assigned to taxonomies using the Ribosomal Database Project alterations associated with fluctuations in Firmicutes-associated (RDP) classifier.24 A confidence value of 0.5 was considered taxa; alterations in the proportions of the Bacteroidetes and a positive identification. At each phylogenetic classification level, e Proteobacteria phyla also featured in several studies.10 15 the overall data from each subject was normalised by scaling to However, in studies to date, the phenotypic characterisation of an intensity of 1 in order to control for differing numbers of the subjects has often been suboptimal. For example, links to reads. To visualise the data, we employed principal coordinates symptom pattern and severity have not been clearly established. analyses (PCoAs) based upon the Unifrac distances25; the Moreover, it is currently unclear if the gut microbiota alterations UniFrac measurement indicates the extent to which microbial found were characteristic of IBS patients in general or only to communities share branch length, based on a phylogenetic tree subgroups thereof. of all the amplicon sequences in the combined dataset. The aim of this study was, therefore, to identify differences in Secondary clustering analysis was carried out on the OTU the intestinal microbiota between healthy and IBS subjects in abundances by adding 0.000001 to remove zeros followed by a very well phenotyped cohort. Using faeces as a proxy for the dual scaling a log transformed OTU dataset to limits of À3,3 and distal bowel, we sought to identify phylum, genus and species using unsupervised hierarchical cluster analysis with the Pearson level bacterial signatures of IBS. We further aimed to identify correlation coefficient and average linkage. All sequence reads are operational taxonomic units (OTUs) associated with IBS and deposited at MG-RAST (Project ID: 152). their relationship to clinical patterns and physiological Subsequent feature selection was carried out using the measures. KruskaleWallis test. For phylum, class, order and family taxo- nomic levels, correction for multiple testing was applied using METHODS Holm’s method.26 For the species and genus level analysis, the Recruitment of subjects less stringent method of Benjamini and Hochberg was used.27 IBS patients fulfilling Rome II criteria16 were recruited from the This method controls the false discovery rate and so is a less Gastroenterology Clinic, Sahlgrenska University Hospital, stringent condition than the control of the family-wise error Gothenburg, Sweden. The samples were collected between rate. When applying statistics, a cut-off of 20% occupancy of January and December 2004 to cover a complete calendar year a taxa was used to remove rare or poorly measured species. This and eliminate seasonal effects. All patients underwent appro- means that a variable was removed from testing if it contained priate diagnostic tests to rule out organic GI disorders as 46 or more zeros. Testing of rates of depression across the described in the online supplementary methods. Patients with IBS groups was carried out using the KruskaleWallis test other GI conditions explaining their symptoms, or with another and confirmed in a pairwise manner using the asymptotic severe coexisting disease, were excluded. Of the 37 patients, 27 WilcoxoneManneWhitney rank sum test. were on no medication during the study; 3 were on a proton pump inhibitor, 2 used spasmolytics, 4 bulking agents, 4 selec- Assigning species and Clostridium clusters tive serotonin reuptake inhibitors, 2 low dose tricyclics and 2 To confidently assign amplicon sequences to species level and loperamide. We also recruited a group of healthy volunteers Clostridium clusters, we extracted 72 928 V4 sequences from the without previous or current GI symptoms or history of a chronic 123 332 full-length 16S rRNA genes with complete species disease. These subjects were recruited through local advertise- classifications (RDP release 10.23) using the same primer pair ment and completed a GI symptom questionnaire to ensure the