Pouchitis is an inflammatory condition which commonly occurs in patients that have undergone a colectomy followed by ileal pouch-anal anastomosis (IPAA). IPAA is the surgical creation of an artificial colon from ileal gut tissue and pouchitis, frequently caused by the microbiome present, is the inflammation of this artificial colon, also known as the ileal pouch. Interestingly, the occurrence of this condition differs between individuals and the illness that led them to IPAA. Therefore, studying and comparing the microbiomes and transcriptomes of IPAA patients will help with our understanding of the causes of pouchitis and hence ways this can be prevented. Recent research published in Genome Biology has investigated this and here we ask co-authors Curtis Huttenhower and Xochitl Morgan from Harvard T. H. Chan School of Public Health, USA, more about their research.
Why did you choose to use the IPAA model of inflammatory bowel disease (IBD) in your study?
IPAA (Ileal Pouch-Anal Anastomosis) is a reconstructive surgery performed after the removal of the colon. In this surgery, a reservoir (the ileal pouch, commonly referred to as a “J-pouch”) is constructed from the end of the small intestine and reconnected to the rectum. About half of ulcerative colitis (UC) patients who have this surgery will have at least one episode of pouchitis, or inflammation in the pouch, and for some of them, it becomes a chronic issue. However, pouchitis is very rare in patients who have this surgery due to colon cancer. We know that pouchitis is more strongly linked to the microbiome than any other type of inflammatory bowel disease because antibiotics are an extremely effective treatment. We also know that genetic factors must influence this process, because ulcerative colitis is a genetically-linked disease (over 130 genes are currently associated with UC risk), and UC patients get pouchitis at about ten times the rate of other patients who have this surgery. Due to this combination of strong microbiome and host genetic influence, IPAA patients are a very interesting model to study the role of host-microbe interactions and try to understand how they contribute to the development of pouchitis. In our study, we studied the microbiomes and host transcriptomes of IPAA patients with UC and patients with familial adenomatous polyposis (FAP), a form of hereditary colon cancer that is caused by mutations in a single gene, in order to try to understand how they are different, and which transcripts are related to the abundance of specific microbes.
Host genetics and the microbiome are known to influence the development of pouchitis, though little is known about how they interact. Why do you think this is, and how does your study address this knowledge gap?
This is a challenging problem to study for various reasons. Firstly, we were handicapped by technology until relatively recently – microbial surveys of large populations were not really logistically or financially feasible for most investigators until after Roche released the GS FLX sequencer in 2008. This allowed large projects like MetaHit (2010) and the Human Microbiome Project (2011). Until then we did not know how much variation to expect in the microbiome of a normal large Western population, let alone a sick population. After MetaHit and the HMP, we were much better informed about how much power was necessary for this kind of study.
Secondly, this is a challenging problem from a data analysis perspective. For this study, we wanted to find correlations between host transcripts and microbes, but there were approximately 20,000 host transcripts and around 7,000 unique bacterial OTUs in our study population. This meant that we had to find novel strategies to be well-powered to find associations in high-dimensional data in our cohort of 200 patients. I hope other investigators who are struggling with the same sorts of statistical problems may find some of these strategies helpful in their own studies. In addition, our study defined major ecological effects on the microbiome and host transcriptome, and knowing these will be very important for researchers.
How did you go about designing your study and what challenges did you face?
Our collaborators, Dr. Mark Silverberg’s group at Mt Sinai Hospital, Canada, collected a beautiful dataset to study the genes and microbes associated with recurrence of inflammation following ileal resection. We could examine tissue effects because many individuals were biopsied in two locations, clinical outcome, or the effects of inflammation or antibiotic use. Every microbiome study presents unique challenges because so many environmental factors can influence the microbial data. In this particular clinical dataset, we had to properly account for the environmental factors of antibiotic use and inflammation, which were common among the patients.
Your use of data reduction methods enabled you to identify the significant associations between host transcripts and microbes. Why did you choose to do this, and how do you think these methods can be used in other microbiome studies?
We used two novel approaches to increase power in high-dimensional data. The first hypothesized that previous work, such as genome-wide association studies of IBD, would determine the most important genes involved in host-microbe interactions. In our second approach, we automatically identified the most variable transcripts and microbes with principal components analysis. We found that the second approach was more effective, suggesting that the data-driven approach was best.
Your study showed that host transcripts vary with location and inflammation, but that the association of transcripts to the microbiome is weak. Were you surprised by this? What do you think are the underlying reasons for this?
We measured the effects of antibiotic use, biopsy location, and inflammation on both the transcriptome and microbiome, and found that the transcriptome was mostly influenced by biopsy location (and secondarily by inflammation), while the microbiome was heavily influenced by antibiotic use and individual variation. Given the large size of these effects, the modest association between host transcripts and microbes was not surprising. However, these results improve our understanding of the major driving forces of microbial community ecology in pouches. In as much as host transcription is much faster than microbial growth, the modest association we observe may also be due to a mismatch in time scale.
Your study provided an in depth look at the effects of antibiotic use on the microbiota of your study participants. What were your main findings?
Metronidazole is the gold standard antibiotic for pouchitis patients, so many of the patients in our cohort were taking this drug. Metronidazole has strong effects on anaerobic gut bacteria, so when patients take it, we see extreme reductions in the Bacteroidetes, and in many of the Firmicutes (classes Clostridiaceae, Lachnospiraceae, Peptostreptococcaceae, and Ruminococcaceae), and increases in resistant clades such as Enterococcus and Enterobacteriaceae.
How can your investigation of the relationship between host transcription and microbial communities be used to inform clinical outcomes? What further research is needed to address this more completely?
We found that using microbiome and transcriptome data, we could mostly discriminate FAP patients and patients with Crohn’s disease-like inflammation, but we could not discriminate between single-episode pouchitis patients and never-pouchitis patients. If the mechanisms underlying the latter are not entirely transcriptional or microbial, they may be epigenetic or immune-mediated, and we must consider these other mechanisms (and consider environmental factors such as antibiotic use). The ideal method to determine which factors predict and influence long-term outcome would be to recruit patients at the time of surgery and collect samples longitudinally so that we could look for temporal changes that preceded the shift from health to disease.
Questions by Barbara Cheifet, Editor of Genome Biology
The post Host-microbe interactions leading to inflammation following colectomy appeared first on Biome.