Two broad approaches have been taken to define the role of the microbiota in IBD: the “global description strategy”126–128 and the “candidate microorganism strategy.”129 The former describes the attempt to characterize the composition of the microbial community and has led to the concept of “dysbiosis.” The “candidate microorganism strategy” refers to the identification of single specific microorganisms that are thought to be directly pathogenic, such as Escherichia coli129 or Mycobacterium avium subspecies paratuberculosis (MAP).130 These two approaches are not mutually exclusive.
“Global Description Strategy”
While dysbiosis is defined broadly as a deviation of the microbial community from its normal state, in IBD a number of consistent features have been identified. The FAM of CD28, 82, 131 and UC127 patients is characterized by the presence of bacteria that do not belong to the usual dominant phylogenetic groups. In the mucosa-associated microbiota (MAM) of CD patients there are increased concentrations of total bacteria,132–137 and total and facultative anaerobes.133, 138, 139 In the MAM of UC patients there are increased concentrations of total bacteria,133–135, 137 total anaerobes,133, 140 and aerobes.141
The definition of the nature and diversity of the microbial spectrum is dependent in part on the techniques used to characterize it. Dysbiosis has been characterized by an overall decrease in biodiversity in the MAM and FAM of CD127, 131, 142–145 and UC143, 145–148 compared with controls. However, recent pyrosequencing studies suggest that the FAM of UC patients in remission is similar to that of healthy controls.91
CD appears to be associated with a reduction in the diversity of the phylum Firmicutes, particularly the Clostridium leptum subgroup, in both the FAM82, 127, 131, 142 and MAM.3, 145, 149 In particular, the species Fecalibacterium prausnitzii, which has antiinflammatory properties,120 is less prevalent in the FAM150, 151 and MAM3, 152, 153 of CD patients, as is the genus Roseburia.91 In UC, Firmicutes, particularly the C. coccoides group, has been shown to be reduced in both FAM127 and MAM.128, 145F. prausnitzii has been shown in a single study to be increased in UC.150
The Ruminococcaceae family appear to be more abundant in the FAM and MAM of healthy subjects, compared with elderly subjects154 and CD,91, 155, 156 and in females compared with males.157 Some Ruminococcus species also appear to be more abundant in the FAM and MAM of active UC.158, 159 The relative abundance of R. gnavus has been found to be increased in ileal CD.91 The pathogenic significance of this microbe is unclear; it has been reported to both produce a bacteriocin with antiinflammatory properties,160 and to have mucolytic properties.161
Using a variety of techniques, increased counts of bacteria, especially E. coli and Enterococci, have been found in the FAM28, 156, 162 and MAM133, 152, 155, 159, 163–165 of both active and inactive CD and UC, compared with healthy controls.
There are conflicting data regarding the phylum Bacteroidetes. In the FAM and MAM of CD and UC versus controls, Bacteroidetes and its species has been found to be both increased128, 131, 133, 144, 145, 151, 166, 167 and decreased.3, 136, 142, 148, 155
Increased numbers of Fusobacteria have been found in the FAM140 and MAM158 of active UC compared with inactive UC and controls.
In CD Bifidobacteria and Lactobacilli have been found to be decreased in the FAM in both adults28, 142, 156, 168 and children.151 In the MAM of UC patients, Bifidobacteria are decreased,164 whereas Lactobacilli have been found to be decreased in the FAM of patients with active UC.169
Recently, TM7 (a subgroup of Gram-positive uncultivable bacteria), previously implicated in oral inflammation, has been shown to be increased in diversity in active CD (23 phylotypes) compared with active UC (10 phylotypes) and non-IBD controls (12 phylotypes). The TM7 associated with CD and UC was strongly associated with antibiotic resistance compared with controls.170
Swidsinski et al150 suggested that two features could be useful as a “fingerprint” to discriminate between CD and UC: a reduction in the concentration of F. prausnitzii in CD to less than 1 × 109 per mL, and an increase in leukocytes in UC to >30 leukocytes/104μm2. Recent studies have also demonstrated differences in the relative abundance of the Bifidobacteriaceae, Coriobacteriaceae, and Ruminococcaceae families among individuals with different CD phenotypes.91, 144, 153
In a study of FAM using DGGE, Joossens et al171 found a “dysbiosis signature” associated with CD, characterized by five bacterial species, namely, a decrease in Dialister invisus, F. prausnitzii, and Bifidobacterium adolescentis, and an increase in R. gnavus and an uncharacterized species of Clostridium cluster XIVa.
In the same study, the fecal samples of 84 unaffected first-degree relatives of CD patients were found to have an altered composition of the predominant microbiota compared with controls. No difference was seen when F. prausnitzii was specifically targeted by qPCR.171 Prospective studies are needed to identify if overt disease will develop in some of these relatives over time.
“Candidate Microorganism Strategy”
The two “candidate organisms” that have received the most attention as having a possible specific association with CD are E. coli and MAP.
MAP has long been considered a possible causative agent in CD because of the similarity of Johne's disease in cattle, caused by MAP, and CD in humans.172
Of two recent meta-analyses and systematic reviews of the association of MAP with CD the first included 28 case-controlled studies of MAP DNA detected by PCR in tissue samples, or antibodies against MAP antigens tested by enzyme-linked immunosorbent assay in serum. MAP was detected more often in patients with CD than in controls and patients with UC.173 The second meta-analysis only included studies of tissue samples using nucleic acid-based techniques, specifically PCR or in situ hybridization. In the analysis of the 47 studies, MAP was detected more frequently among CD patients compared with controls.174 This meta-analysis has been criticized on its exclusion of 13 studies in which MAP was not detected in any patients with CD or in controls.175 In a recent large-scale 16S rRNA gene library study which was not included in either of these two meta-analyses, more than 15,000 small subunit ribosomal RNA genes were analyzed and MAP was not detected in CD patients.3
The influence of IBD medication on MAP should be considered. Methotrexate, 6-mercaptopurine, and 5-aminosalicylic acid have been shown to inhibit MAP growth in vitro.176, 177 Kirkwood et al130 investigated pediatric patients not yet treated with any medication and found that MAP was identified more often in mucosal biopsies and peripheral blood mononuclear cells from CD than in non-IBD patients. Difficulties in extraction of MAP DNA may explain the failure and detection in many studies.
The effect of antimycobacterial therapy has been the subject of a Cochrane review, which cautiously concluded that antimycobacterial therapy may have a role in maintaining remission in CD, but fell short of recommending therapy.178 The result of the largest randomized controlled trial to date does not support the use of antimycobacterial therapy in CD.179
Numerous studies have shown increased numbers of mucosa associated E. coli in both CD and UC compared with healthy controls.129, 132, 133, 136, 153, 163, 164, 180–182 As a commensal organism within the normal gut microbiota, E. coli plays an important role in maintaining intestinal homeostasis and is not implicated in disease unless there is a breach in the intestinal mucosa barrier. Of the colonic-like microbiota that colonize the neoterminal ileum postresectional surgery for CD, E. coli tend to predominate.180, 181
The adherent invasive strain of E. coli (AIEC) has developed virulence factors that allow it to adapt and survive in the postoperative environment.183 AIEC has been found in association with early neoterminal ileal lesions in the postoperative CD setting.165 The ability of AIEC to adhere and invade intestinal cells is mediated by a number of virulence factors including type 1 pili, flagella, and outer membrane porin C.184–188 Moreover, AIEC is able to resist phagocytosis, and survive and replicate extensively in large vacuoles within macrophages without triggering host cell death.189–191
The type 1 pili of AIEC bind to the specific receptor CEACAM-6 expressed in ileal epithelial cells of patients with CD but not healthy controls.192 CEACAM-6 receptors become overexpressed in response to stimulation of ileal epithelial cells by tumor necrosis factor alpha (TNFα), which is released from macrophages that have taken up CD-associated AIEC.193 AIEC therefore causes an amplification loop of colonization and inflammation.190 The early inflammatory response to AIEC among patients with CD carrying CARD15 polymorphisms appears to be disturbed.194
Several other bacteria such as Pseudomonas,195–199Yersinia,200, 201Listeria,155, 202–204Burkholderia,205 and Helicobacter206–210 have been linked to IBD, but what pathophysiological role, if any, they play remains to be determined. A contribution from viruses211–213 and fungi214–216 has also not been excluded.