Introduction
The human microbiome is a complex ecosystem of bacteria, archaea, viruses, and eukarya found virtually along every surface of the human body (Shreiner et al., Reference Shreiner, Kao and Young2015; Berg et al., Reference Berg, Rybakova, Fischer, Cernava, Vergès and Charles2020; Ferrie et al., Reference Ferrie, Webster, Wu, Tan and Carey2020). Microbiomes are key contributors to health and disease via important host–microbiota interactions (Shreiner et al., Reference Shreiner, Kao and Young2015). The recent introduction of culture-independent analytical techniques, from metagenomics to metabolomics has made detailed study of the microbiome possible (Shreiner et al., Reference Shreiner, Kao and Young2015).
The gut microbiome is crucial for intestinal health maintenance, and its role in nutrition-related, metabolic and inflammatory disorders has previously been established (Doré et al., Reference Doré, Simrén, Buttle and Guarner2013; Shreiner et al., Reference Shreiner, Kao and Young2015; Ferrie et al., Reference Ferrie, Webster, Wu, Tan and Carey2020). Diversity and richness of microbiome increases with distal progression along the gastrointestinal tract (GIT), although this varies greatly between and within individuals (Shreiner et al., Reference Shreiner, Kao and Young2015; Ferrie et al., Reference Ferrie, Webster, Wu, Tan and Carey2020). The volume of colonic microbiota exceeds that of all other organs by at least two orders of magnitude (Sender et al., Reference Sender, Fuchs and Milo2016) and is chiefly implicated in discussions concerning the “gut microbiome” (Shreiner et al., Reference Shreiner, Kao and Young2015).
The gut microbiome is sensitive to environmental changes such as diet, smoking, antibiotics, and even gastrointestinal surgery (Shreiner et al., Reference Shreiner, Kao and Young2015; Rolhion and Chassaing, Reference Rolhion and Chassaing2016; Valdes et al., Reference Valdes, Walter, Segal and Spector2018; Ferrie et al., Reference Ferrie, Webster, Wu, Tan and Carey2020). In this context, maintenance in microbiome diversity may protect against these changes by providing stability, with a reduction in diversity often associated with pathological conditions such as inflammatory bowel disease and infectious colitis in the case of C. difficile (Ferrie et al., Reference Ferrie, Webster, Wu, Tan and Carey2020).
Recent large trials such as the Rotterdam Study (RSIII) approximated the colonic microbiome via faecal studies, showing the dominant phyla as Bacillota (77.8%) and Bacteroidia (12.5%), with lesser extents of Pseudomonadota (4.9%) and Actinomycetota (4.1%). These findings are consistent with other similar studies (Zhernakova et al., Reference Zhernakova, Kurilshikov, Bonder, Tigchelaar, Schirmer and Vatanen2016; Deschasaux et al., Reference Deschasaux, Bouter, Prodan, Levin, Groen and Herrema2018). Recent results from the Dutch Microbiome Project have suggested that individual environmental factors contribute significantly to the interindividual variability of the microbiome (Gacesa et al., Reference Gacesa, Kurilshikov, Vich Vila, Sinha, Klaassen, Bolte, Andreu-Sánchez, Chen, Collij, Hu, Dekens, Lenters, Björd, Swarte, Swertz, Jansen, Gelderloos-Arends, Jankipersadsing, Hofker and Weersma2022).
No singular definition of a healthy microbiota exists, due in part to the heterogeneity of existing studies, but also because of the huge variance within the human microbiome, which has yet to be fully accounted for (Lightner and Pemberton, Reference Lightner and Pemberton2017). One way to define health, as seen with the Dutch Microbiome Project, is to correlate patterns of bacterial presence, recognised as “signatures” of health, with disease and medication use (Gacesa et al., Reference Gacesa, Kurilshikov, Vich Vila, Sinha, Klaassen, Bolte, Andreu-Sánchez, Chen, Collij, Hu, Dekens, Lenters, Björd, Swarte, Swertz, Jansen, Gelderloos-Arends, Jankipersadsing, Hofker and Weersma2022). In other studies, low levels of specific bacteria such as Pseudomonadota combined with abundance of signature SCFA-producing genera from the other three phyla such as Bacteroidia, Ruminococcus, Lactobacillus and Bifidobacterium generally indicate a functional colonic environment in homeostasis (Shreiner et al., Reference Shreiner, Kao and Young2015).
Faecal diversion involves creation of an ostomy (typically ileostomy or colostomy) to divert the faecal stream from the distal end of the GIT (Remzi, Reference Remzi2017). This is most commonly performed following a low anterior resection for rectal cancer, particularly after radiotherapy, or acute colonic resections where inflammation or infection increases the risk of anastomotic leak,. Faecal diversion can be temporary or permanent, and is designed to mitigate the risk of severe sepsis in the event of an anastomotic leak. Diversion without resection may also be performed in severe perianal fistulising disease to promote perianal healing by preventing lesion-to-stool contact, such as in Crohn’s disease (CD) (Whelan et al., Reference Whelan, Abramson, Kim and Hashmi1994; Remzi, Reference Remzi2017).
The stoma results in a functional end that receives nutrients from the faecal stream and a defunctioned end which does not. The diverted or defunctioned end is at high risk of diversion colitis (DC) (~70-90% by various estimates) (Ten Hove et al., Reference Ten Hove, Bogaerts, Bak, Laclé, Meij and Derikx2018; Pieniowski et al., Reference Pieniowski, Nordenvall, Palmer, Johar, Tumlin Ekelund and Lagergren2020). Treatment may involve stoma reversal, which often improves symptoms; however, these patients are then at increased risk of developing lower anterior resection syndrome (LARS) and C. difficile colitis (~18-55%, and ~1-4%, respectively) (Harries et al., Reference Harries, Ansell, Codd and Williams2017; Dou et al., Reference Dou, Gao, Yan and Shan2020). While the precise pathophysiology is unclear, limited preliminary evidence suggests that colonic microbiome alterations due to diversion may be a contributing factor.
In murine models, oral short-chain fatty acids (SCFA) has been used successfully to treat various forms of murine colitis via restoration of gut microbiota–host interactions (Harig et al., Reference Harig, Soergel, Komorowski and Wood1989). In humans, faecal microbial transplants (FMTs) have also been effectively employed to treat recurrent C. difficile colitis while SCFA enemas have also shown limited success at reducing symptoms of DC patients (Rao and Safdar, Reference Rao and Safdar2015; Radjabzadeh et al., Reference Radjabzadeh, Boer, Beth, van der Wal, Kiefte-De Jong and Jansen2020). We therefore hypothesise that loss of enteral nutrition in the diverted colon results in dysbiosis, especially of SCFA-producing microorganisms, consequently impacting intestinal structure, function and immunity leading to increased risk of inflammation and disease. Understanding the colonic microbiome changes that occur in the context of diversion may thus be key in characterising and managing these adverse outcomes.
Given the potential relevance of the microbiome in dysbiosis outcomes following diversion, there is a need to understand microbiome changes in the diverted colon. Recent studies are few and heterogenous. Therefore, we seek to systematically review the existing literature, identify key knowledge gaps and highlight areas requiring further attention.
Our aims are to: firstly, characterise the longitudinal changes in the colonic microbiome that occur post-diversion, and secondly, to identify microbiome characteristics associated with dysbiosis related outcomes post-diversion.
Methodology
Search strategy
A systematic search was designed according to PRISMA guidelines (Figure 1). The search strategy involved searching combinations of keywords and MeSH terms related to 2 key concepts – diversion and microbiome – in the MEDLINE, EMBASE and CENTRAL databases. An example of the MEDLINE search is shown in Figure 2 and adapted as required for EMBASE and CENTRAL, respectively. We also performed secondary backward and forward citation searching on all included papers as well as potentially relevant reviews. Two independent reviewers conducted screening, inclusion and data extraction, with disputes settled by discussion or with a third independent reviewer if consensus was not reached.
Figure 1. Prisma diagram.
Figure 2. Search strategy for MEDLINE (22 Jul 2022).
Inclusion and appraisal
We included studies involving adult participants >18y that underwent faecal stream diversion – defined as an ileostomy or colostomy – with measured outcomes that included microbiome analysis on the defunctioned colon post-diversion. Studies that examined microbiome differences pre- and post-diversion, or between functional and defunctioned mucosa in the same individuals were included, as well as those that utilised external controls. While this was not ideal, we believe conclusions within the studies were still informative and valid given the broadly identifiable microbiome trends in healthy external controls.
Paediatric populations were excluded due to their different microbiome composition (Joanna Briggs Institute, 2022). Diversion above the jejunum such as biliopancreatic diversion was also excluded as these procedures are not typically associated with colonic dysbiosis outcomes investigated here. Animal-related, non-English, non-full text articles and studies preceding 1998 were also excluded.
Quality and bias assessment was subsequently done on all included papers using the JBI Appraisal tool (Joanna Briggs Institute, 2022). Using the JBI tool, a scoring system similar to Ferrie et al. was used (Ferrie et al., Reference Ferrie, Webster, Wu, Tan and Carey2020). 1 point was assigned for “Yes” or “NA,” 0 points for “No” and 0.5 for “Unclear” (Table 1).
Table 1. Characteristics of studies: Population demographics
Table 1 showing population demographics for included studies. Only relevant data extracted. In studies investigating multiple interventions or multiple comparison groups, only data directly pertaining to the impact of diversion was extracted. In Watanabe et. al. for example, both case and control groups were reported as both groups received faecal diversion.
Nb: Tominaga et al. (Reference Tominaga, Tsuchiya, Mizusawa, Matsumoto, Minemura, Oka, Takahashi, Yoshida, Kojima, Ogawa, Kawata, Nakajima, Kimura, Abe, Setsu, Takahashi, Sato, Ikarashi, Hayashi, Mizuno, Yokoyama, Tajima, Nakano, Shimada, Kameyama, Wakai and Terai2021a, Reference Tominaga, Tsuchiya, Mizusawa, Matsumoto, Minemura and Oka2021b) had an overlap of 5 similar patients. Both were, however, different studies altogether – differing in comparisons used, outcomes measured etc. (See Table 2 for full detail) Both were therefore included and treated separately.
CA, cancer; CD, Crohn’s disease; IPAA, ileal pouch anal anastomosis; UC, ulcerative colitis.
1 Control group numbers not added to sample size unless explicitly stated. Only patients who underwent faecal diversion included.
2 JBI tools do not include an NRS checklist, so RCT checklist was used and NA (1 point) assigned to non-applicable criteria.
Study selection
The summary process and exclusion reasons are shown in full in the PRISMA diagram (Figure 1). The review is reported in keeping with PRISMA guidelines.
Results
Included studies
The primary search was conducted on 22/7/22 and identified 738 records after duplicate removal. Following title and abstract screening, Forty-two articles were appraised in full. Five additional articles were further identified during secondary searching and appraised. Six articles were included in the final review.
Characteristics of included studies
We included 3 case–control, 2 cohort and 1 non-randomised controlled study involving 95 (47m:48f) patients in total, who were generally older in age (>55y) (Table 1). Broadly speaking, most of the studies were small (n<35) and involved diversion procedures in relation to malignancy or IBD. Most of the patients sampled underwent loop ileostomies (n=82), while others had loop (n=10) and end (n=3) colostomies. Apart from this, the studies were heterogenous with regards to sampling and analysis methods, as well as comparators (Table 2). Three studies each utilised external and internal controls, respectively. External controls included healthy patients or patients who underwent non-diversion surgery; internal controls consisted of mucosa comparisons between diverted and proximal colons (singe time point) or longitudinal sampling of the colon in relation to faecal diversion or restoration, which provided temporal data. All studies, except Young et al. (Reference Young, Raffals, Huse, Vital, Dai, Schloss, Brulc, Antonopoulos, Arrieta, Kwon, Reddy, Hubert, Grim, Vineis, Dalal, Morrison, Eren, Meyer, Schmidt, Tiedje, Chang and Sogin2013), Baek et al. (Reference Baek, Kim, Lee, Roh, Keum, Kim and Kim2014), Beamish et al. (Reference Beamish, Johnson, Shaw, Scott, Bhowmick and Rigby2017), Tominaga et al. (Reference Tominaga, Tsuchiya, Mizusawa, Matsumoto, Minemura, Oka, Takahashi, Yoshida, Kojima, Ogawa, Kawata, Nakajima, Kimura, Abe, Setsu, Takahashi, Sato, Ikarashi, Hayashi, Mizuno, Yokoyama, Tajima, Nakano, Shimada, Kameyama, Wakai and Terai2021a, Reference Tominaga, Tsuchiya, Mizusawa, Matsumoto, Minemura and Oka2021b), Watanabe et al. (Reference Watanabe, Mizushima, Okumura, Fujino, Ogino and Miyoshi2021), sampled mucosal biopsies (among other methods) which are generally considered more representative of the mucosal microbiome. However, microbiome analysis methods differed with Beamish et al. (Reference Beamish, Johnson, Shaw, Scott, Bhowmick and Rigby2017) and Baek et al. (Reference Baek, Kim, Lee, Roh, Keum, Kim and Kim2014) opting for PCR or culture-dependent methods instead of gene sequencing. It is also worth noting some studies such as Watanabe et al. ((Reference Watanabe, Mizushima, Okumura, Fujino, Ogino and Miyoshi2021) included other forms of dysbiosis measures such as histopathology and cytometry which provide information regarding intestinal health in addition to microbiome changes. Sample times varied significantly ranging from 1 to 40 months post-diversion. In summary, the studies included were generally small and heterogenous; therefore, a meta-analysis was not possible; we opted instead to perform a qualitative and narrative synthesis of the available data. Methods of DNA extraction, sequencing and analysis methods are summarised in Table 3.
Table 2. Characteristics of studies: Methods
Table 2 summarising the methodology of included studies revealing significant heterogeneity between studies. Only relevant data pertaining to bowel diversion was extracted.
CD, Crohn’s disease; CRC, colorectal cancer; DC, diversion colitis; FC, flow cytometry; IHC, immunohistochemistry; N/S, not stated.
Table 3. Summary of methods used for specimen collection, DNA extraction, sequencing and analysis
Diversion and diversity
Five of the included 6 studies (excluding Baek et al., Reference Baek, Kim, Lee, Roh, Keum, Kim and Kim2014) reported on diversity measures with results summarised in Table 4. Beamish et al. (Reference Beamish, Johnson, Shaw, Scott, Bhowmick and Rigby2017) reported a reduction in total mucosal bacterial load (-62.4%) and DGGE band profiling (-5 bands) between diverted and proximal mucosa. Tominaga et al. (Reference Tominaga, Tsuchiya, Mizusawa, Matsumoto, Minemura, Oka, Takahashi, Yoshida, Kojima, Ogawa, Kawata, Nakajima, Kimura, Abe, Setsu, Takahashi, Sato, Ikarashi, Hayashi, Mizuno, Yokoyama, Tajima, Nakano, Shimada, Kameyama, Wakai and Terai2021a) reported a decrease in alpha diversity (chao1 p<0.01, OTU p<0.01) and significant difference in beta diversity (Unifrac p<0.01) when comparing mucosal microbiome of DC patients to faeces of healthy controls. The mucosal–faecal comparison was not ideal; however, the authors presumably wanted to avoid subjecting healthy controls to unnecessary biopsies. Moreover, the findings of decreased alpha and beta diversity are consistent with other included studies such as Watanabe et al. (Reference Watanabe, Mizushima, Okumura, Fujino, Ogino and Miyoshi2021) which add validity. Interestingly, Tominaga et al. (Reference Tominaga, Tsuchiya, Mizusawa, Matsumoto, Minemura and Oka2021b) who, like Beamish et al. (Reference Beamish, Johnson, Shaw, Scott, Bhowmick and Rigby2017) compared microbiome compositions of the proximal and diverted colon reported no difference in alpha diversity (chao1 p=0.69, Shannon p=0.23) although the difference in beta diversity was significant (Unifrac p<0.05), signifying a difference in microbiome composition. It is worth noting however, that unlike Beamish et al. (Reference Beamish, Johnson, Shaw, Scott, Bhowmick and Rigby2017) (n=34), this study was much smaller (n=8) and therefore may simply have been underpowered. Young et al. (Reference Young, Raffals, Huse, Vital, Dai, Schloss, Brulc, Antonopoulos, Arrieta, Kwon, Reddy, Hubert, Grim, Vineis, Dalal, Morrison, Eren, Meyer, Schmidt, Tiedje, Chang and Sogin2013) and Watanabe et al. (Reference Watanabe, Mizushima, Okumura, Fujino, Ogino and Miyoshi2021) were the only two studies utilising longitudinal sampling. The former found significantly decreased alpha diversity and reductions in viable cell counts in the diverted mucosa prior to ileostomy reversal compared to after. Surprisingly, the alpha diversity increased to the range of healthy control samples after 2 months post-reversal; however, the viable cell counts, though increased, remained lower than controls. Watanabe et al. (Reference Watanabe, Mizushima, Okumura, Fujino, Ogino and Miyoshi2021) similarly concluded that alpha diversity (chao1 p=0.001, Shannon p<0.001, OTU p=0.015) and mucosal bacterial load (p<0.01) were markedly reduced in the diverted colon compared to the proximal colon. Post-reversal analysis was not done with respect to the microbiome. In conclusion, there is limited but significant evidence that diversion and enteral nutrient deprivation reduces microbiome diversity, possibly predisposing patients to dysbiosis-related outcomes.
Table 4. Diversity changes following diversion
Table 4 summarising microbiome diversity changes following diversion. Only significant increases or decreases reported. P values recorded together with diversity measure if available. Pre- and post-ileostomy reversal data reported where available.
DGGE, denaturing gradient gel electrophoresis; N/S, not stated; OTU, operational taxonomic unit.
Phylum- and genus-specific changes
Five out of 6 studies (excluding Tominaga et al., Reference Tominaga, Tsuchiya, Mizusawa, Matsumoto, Minemura, Oka, Takahashi, Yoshida, Kojima, Ogawa, Kawata, Nakajima, Kimura, Abe, Setsu, Takahashi, Sato, Ikarashi, Hayashi, Mizuno, Yokoyama, Tajima, Nakano, Shimada, Kameyama, Wakai and Terai2021a) reported phylum- and genus-specific changes, which are summarised in Table 5. Increases or decreases were defined as at least one supporting study without any conflict.
Table 5. Effect of diversion on genus/phylum composition of the gut microbiome
Table 5 highlighting genus/phyla specific changes associated with diversion together with their potential significance. Increases/Decreases were defined as at least 1 supporting study among the included studies without any conflict with the others. Unclear was defined as conflicting data within a single study or between studies. Only significant results reported.
Table format and analysis adapted from Ferrie et al. (Reference Ferrie, Webster, Wu, Tan and Carey2020).
At a phylum level, both Beamish et al. (Reference Beamish, Johnson, Shaw, Scott, Bhowmick and Rigby2017) and Young et al. (Reference Young, Raffals, Huse, Vital, Dai, Schloss, Brulc, Antonopoulos, Arrieta, Kwon, Reddy, Hubert, Grim, Vineis, Dalal, Morrison, Eren, Meyer, Schmidt, Tiedje, Chang and Sogin2013) reported significantly decreased (21% reduction, p= 0.02) Bacillota compositions in the diverted colon. The former also reported an increase in Pseudomonadota, (6.9%, p=0.05) but found significant variation in Bacteroidia composition, as opposed to Young et al. (Reference Young, Raffals, Huse, Vital, Dai, Schloss, Brulc, Antonopoulos, Arrieta, Kwon, Reddy, Hubert, Grim, Vineis, Dalal, Morrison, Eren, Meyer, Schmidt, Tiedje, Chang and Sogin2013) who described a decrease in Bacteroidia. While the significance of Bacteroidia and Bacillota changes are unclear without species-specific information, Pseudomonadota increases are strongly indicative of a microbial “dysbiosis signature” due to its abundance of pathogenic genera (Shin et al., Reference Shin, Whon and Bae2015). Thus, Bacillota reduction with concomitant increases in Pseudomonadota suggests dysbiosis in the diverted colon (Shin et al., Reference Shin, Whon and Bae2015).
At a genus level, Baek et al. (Reference Baek, Kim, Lee, Roh, Keum, Kim and Kim2014) found diversion correlated with decreases in Lactobacillus (p=0.038) and Bifidobacterium (p<0.001), with Tominaga et al. (Reference Tominaga, Tsuchiya, Mizusawa, Matsumoto, Minemura and Oka2021b) also finding a Lactobacillus decline (p<0.05). Both bacteria are regarded as beneficial due to their roles in metabolism, intestinal immunity and epithelial maintenance, with decreases associated with dysbiosis. Könönen and Wade (Reference Könönen and Wade2015) Interestingly, Baek et al. (Reference Baek, Kim, Lee, Roh, Keum, Kim and Kim2014) additionally found Bifidobacterium as the only genus significantly and inversely correlated with the severity of diversion colitis in patients, highlighting its potential clinical significance. Beamish et al. (Reference Beamish, Johnson, Shaw, Scott, Bhowmick and Rigby2017) also reported a 36.3% decrease in Clostridium abundance in the diverted colon. While the significance of this is debatable without species-level information (as some can be pathogenic), Clostridia are nevertheless a predominant cluster of gut commensals and are generally SCFA-producing (Guo et al., Reference Guo, Zhang, Ma and He2020). Such a large decrease inevitably affects microbiome homeostasis, potentially leading to dysbiosis in the diverted colon.
Other less-specific changes in the diverted colon were also recorded. Tominaga et al. (Reference Tominaga, Tsuchiya, Mizusawa, Matsumoto, Minemura and Oka2021b) found increased levels of Corynebacterium (p<0.01), Peptoniphilus (p<0.05), Anaerococcus (p<0.05) and Porphyromonas (p<0.01). These genera have been associated with haematogenous or tissue infections when translocated or overgrown (Brown et al., Reference Brown, Church, Lynch and Gregson2014; Tidjani Alou et al., Reference Tidjani Alou, Khelaifia, Michelle, Andrieu, Armstrong and Bittar2016; Sędzikowska and Szablewski, Reference Sędzikowska and Szablewski2021; Štšepetova et al., Reference Štšepetova, Simre, Tagoma, Uibo, Peet and Siljander2022). However, their role in intestinal inflammation is less clear at this stage. Porphyromonas for example has been linked to oral infections which does not necessarily translate to intestinal inflammation (Sędzikowska and Szablewski, Reference Sędzikowska and Szablewski2021). Other non-specific changes reported in the diverted colon include decreases in Escherichia (-9%), Streptococci (-36.3%) and increases in Spirosoma (+27%) as reported by Beamish et al. (Reference Beamish, Johnson, Shaw, Scott, Bhowmick and Rigby2017); as well as decreases in Klebsiella (p<0.001), Pseudomonas (p<0.015), Enterococci (p<0.001) and Staphylococci (p<0.038) by Baek et al. (Reference Baek, Kim, Lee, Roh, Keum, Kim and Kim2014) Most of these bacteria are potential pathobionts and these changes when viewed simplistically, could be seen as a positive outcome from diversion (Zhang et al., Reference Zhang, Li, Gan, Zhou, Xu and Li2015; Pettigrew et al., Reference Pettigrew, Gent, Kong, Halpin, Pineles and Harris2018; Martinson and Walk, Reference Ten Hove, Bogaerts, Bak, Laclé, Meij and Derikx2020; Wu et al., Reference Wu, Xu, Su, Li, Lv and Liu2020; Raineri et al., Reference Raineri, Altulea and van Dijl2021; Xi et al., Reference Xi, Song, Han and Qin2021; Roux et al., Reference Roux, Nicolas, Valence, Siekaniec, Chuat and Nicolas2022). However, stability and balance of the microbiome appear to be more important determinants of gut health compared to the absence of specific pathogenic genera.
Following ileostomy reversal, Young et al. (Reference Young, Raffals, Huse, Vital, Dai, Schloss, Brulc, Antonopoulos, Arrieta, Kwon, Reddy, Hubert, Grim, Vineis, Dalal, Morrison, Eren, Meyer, Schmidt, Tiedje, Chang and Sogin2013) reported increases in butyrate metabolism and potentially beneficial SCFA-producing microorganisms such as Acidaminococcus and Coprococcus up to 60 days post-operatively, indicating some level of reversibility; however, the overall microbiome profiles remained different, and its significance was not explored further in these studies.
Other dysbiosis changes
Diversion and microbiome changes were also associated with immunological and functional dysregulation. Both Beamish et al. (Reference Beamish, Johnson, Shaw, Scott, Bhowmick and Rigby2017) (p=0.0004, p=0.01) and Watanabe et al. (Reference Watanabe, Mizushima, Okumura, Fujino, Ogino and Miyoshi2021) (p<0.01, p<0.01) found villous atrophy and reduced crypt cell proliferation in the diverted colon. In addition, Watanabe et al. (Reference Watanabe, Mizushima, Okumura, Fujino, Ogino and Miyoshi2021) also found decreased CD3+ (p=0.037), IL17+ (p=0.002) and IFN-G+ (p=0.013) T-cells signifying immune dysregulation. More importantly, Tominaga et al. (Reference Tominaga, Tsuchiya, Mizusawa, Matsumoto, Minemura and Oka2021b) found decreased SCFA levels (p<0.05) in the diverted colon which adds further weight that the intestinal environment is lacking SCFA-producing bacteria. Interestingly, following ileostomy reversal, Watanabe et al. (Reference Watanabe, Mizushima, Okumura, Fujino, Ogino and Miyoshi2021) reported restoration in villous height, goblet cells and immune cells back to functional ileum levels indicating reversibility of these changes post-ileostomy reversal.
Quality assessment
The quality of papers ranged from scores of 6.5/10 to 9.5/10, and limitations were generally due to sampling methodology or suboptimal controls and outcome measurement. Importantly, different studies also used different methods to characterise the gut microbiota. Baek et al. used quantitative PCR and Beamish used DGGE, while Young, Watanabe and Tominaga (a and b) used 16S rRNA sequencing (Young et al., Reference Young, Raffals, Huse, Vital, Dai, Schloss, Brulc, Antonopoulos, Arrieta, Kwon, Reddy, Hubert, Grim, Vineis, Dalal, Morrison, Eren, Meyer, Schmidt, Tiedje, Chang and Sogin2013; Baek et al., Reference Baek, Kim, Lee, Roh, Keum, Kim and Kim2014; Beamish et al., Reference Beamish, Johnson, Shaw, Scott, Bhowmick and Rigby2017; Watanabe et al., Reference Watanabe, Mizushima, Okumura, Fujino, Ogino and Miyoshi2021; Tominaga et al., Reference Tominaga, Tsuchiya, Mizusawa, Matsumoto, Minemura, Oka, Takahashi, Yoshida, Kojima, Ogawa, Kawata, Nakajima, Kimura, Abe, Setsu, Takahashi, Sato, Ikarashi, Hayashi, Mizuno, Yokoyama, Tajima, Nakano, Shimada, Kameyama, Wakai and Terai2021a, Reference Tominaga, Tsuchiya, Mizusawa, Matsumoto, Minemura and Oka2021b). Additionally, within the 16S rRNA-based studies, different regions were sequenced (Young (V3-V5), Watanabe (V1-V2), Tominaga (a and b) V3-V4), which can lead to technical compositional biases between studies.
A further limitation of these studies was the lack of longitudinal assessment, which could mean that the influence of underlying disease (i.e. malignancy or IBD) causing dysbiosis may not be fully characterised.
Discussion
This systematic review included 6 studies that examined the impacts of faecal diversion on the diverted gut microbiome.
Our review suggests that faecal diversion is associated with a decrease in microbiome diversity, as well as microbiome and intestinal changes suggesting dysbiosis and dysfunction. The loss of diversity together with SCFA-producing bacteria is consistent with inflammatory states such as in IBD as confirmed by other studies (Ferrie et al., Reference Ferrie, Webster, Wu, Tan and Carey2020). While the direction of cause and effect is not immediately clear, a recent RCT found that probiotic stimulation of the diverted bowel loops significantly improved clinical and histological signs of severe and moderate DC in 100% and 88% of patients, respectively (Rodríguez-Padilla et al., Reference Rodríguez-Padilla, Morales-Martín, Pérez-Quintero, Rada-Morgades, Gómez-Salgado and Ruiz-Frutos2021). This points towards an active role of the microbiome in modulating immunity and function rather than simply being a biomarker of dysbiosis. More importantly, this study successfully utilised Lactobacillus and Bifidobacterium in its probiotic therapy, which is again consistent with our findings of decreased Lactobacillus and Bifidobacterium in DC patients. Similarly, Tominaga et al. (Reference Tominaga, Tsuchiya, Mizusawa, Matsumoto, Minemura, Oka, Takahashi, Yoshida, Kojima, Ogawa, Kawata, Nakajima, Kimura, Abe, Setsu, Takahashi, Sato, Ikarashi, Hayashi, Mizuno, Yokoyama, Tajima, Nakano, Shimada, Kameyama, Wakai and Terai2021a) performed autologous FMT on 5 patients with severe DC, achieving 100% subsequent remission. Both study authors conclude that the use of probiotics and FMT, respectively, are effective and safe treatments for DC indicating the importance of the microbiome in the development of future treatments. While our study reaffirmed the partial reversibility of microbiome and intestinal changes following the reversal of faecal diversion, thus supporting it as a treatment for DC; these experimental treatments open future possibilities for treatment options of persistent DC or where reversal is contraindicated.
While the treatment of DC with microbiome modulation has not yet been truly investigated, recent studies show promising results when dealing with inflammatory bowel disease (IBD). Recently, an international Rome consensus was published, acknowledging the role of the gut microbiome in the development of IBD and the utility of faecal microbiota transplant (FMT) as a viable treatment option for mild-to-moderate ulcerative colitis (UC) on a case-by-case basis, although this has not been proven in Crohn’s disease (CD) (Lopetuso et al., Reference Lopetuso, Deleu, Godny, Petito, Puca, Facciotti, Sokol, Ianiro, Masucci, Abreu, Dotan, Costello, Hart, Iqbal, Paramsothy, Sanguinetti, Danese, Tilg, Cominelli, Pizarro, Armuzzi, Cammarota, Gasbarrini, Vermeire and Scaldaferri2023). Instead, assessments of the microbiome could be used to monitor disease activity.
Use of probiotics and prebiotics has also been mooted as a potential longer-term mechanism of modulating the microbiome. The theory of replenishing organisms which are lacking in a specific disease has shown promise in the treatment of IBD (Gowen et al., Reference Gowen, Gamal, Di Martino, McCormick and Ghannoum2023). However, the specific dose and species of probiotic bacterium used to treat a specific condition have yet to be established and remain an area of ongoing research (Lopetuso et al., Reference Lopetuso, Deleu, Godny, Petito, Puca, Facciotti, Sokol, Ianiro, Masucci, Abreu, Dotan, Costello, Hart, Iqbal, Paramsothy, Sanguinetti, Danese, Tilg, Cominelli, Pizarro, Armuzzi, Cammarota, Gasbarrini, Vermeire and Scaldaferri2023).
Our study results are also relevant in the context of prolonged enteral starvation and loop stoma reversal. Ralls et al. found, for example, that total parental nutrition (TPN) use was associated with decreased microbial diversity in humans, as well as a decrease and increase in Bacillota and Pseudomonadota, respectively, similar to the findings by Ralls et al. (Reference Ralls, Miyasaka and Teitelbaum2013), Beamish et al. (Reference Beamish, Johnson, Shaw, Scott, Bhowmick and Rigby2017). These changes were exaggerated with prolonged TPN, and associated with increased anastomotic post-operative leakage and infection (Ralls et al., Reference Ralls, Miyasaka and Teitelbaum2013). The potential link between anastomotic leakage and microbiome dysbiosis is important as it raises the question of whether enteral supplementation of the affected colon (probiotic, SCFA, faecal), especially in the cases of prolonged diversion or TPN, prior to reversal would reduce risk of anastomotic leakage as proposed by Beamish et al. (Reference Beamish, Johnson, Shaw, Scott, Bhowmick and Rigby2017). The results of our study indeed point towards dysbiosis, dysfunction and inflammation that predisposes leakage, thus supporting this recommendation for future trials.
Limitations of this review include the small number of relevant studies and the heterogeneity in methodology between them. Even though we attempted to control for the quality of included studies, the heterogeneity in comparators, sampling time and analysis techniques limits comparison between studies. Previous reviews have shown, for example, that faecal samples can significantly differ from mucosal samples (Jandhyala, Reference Jandhyala2015). Moreover, the sample timings also differed significantly between studies, raising the possibility that some patients may not have been given enough time for the diverted microbiome communities to stabilise before sampling. Bowel preparation use was also unclear in 3 of the 6 studies.
Furthermore, a recent systematic review concluded that even sequencing kits and sample storage conditions may affect microbiome composition (Ferrie et al., Reference Ferrie, Webster, Wu, Tan and Carey2020). Storage temperature, for example, may alter the sequenced Bacteroidia: Bacillota ratio (Ferrie et al., Reference Ferrie, Webster, Wu, Tan and Carey2020). Given the heterogeneity of the included studies and the difficulty of pooled analysis, a standardised protocol for sampling methods, sites and timing, as well as analysis methods in the context of future, adequately powered observational studies or trials are required as recommended by Ferrie et al. (Reference Ferrie, Webster, Wu, Tan and Carey2020).
It is important to note that other components of the gastrointestinal tract such as the virome and mycobiome, which are outside the scope of this review, also contribute to the diversity, health and function of the colon.
Conclusion
This systematic review identified 6 relevant papers examining the impacts of faecal diversion on the diverted gut microbiome. Five of these papers reported significant decreases in microbial diversity after diversion, in addition to phyla and genus-specific changes such as a loss of SCFA-producing genera that support a dysbiosis profile. Moreover, additional immunological and histological evidence support a dysregulated and dysfunctional intestinal environment associated with the microbiome changes.
Restoration of the faecal stream was then associated with improvements in the dysbiosis profile and intestinal function. Novel techniques such as probiotic stimulation and FMT in the efferent limb have shown promise in the treatment of dysbiosis outcomes such as DC. Furthermore, in the context of prolonged starvation or deprivation as is the case for diversion, supplementation of the affected colon may, in theory, reduce anastomotic leakage prior to loop stoma reversal or other forms of reconstructive bowel surgery. These techniques may even have a role to play in improving functional outcomes, but require further investigation to determine their roles and clinical applicability.
The greatest limitations of these studies appear to be scale and power, as the processes involved in sample collection and analysis can be complex and require specific technical ability, due to the high level of interindividual variability and diversity within the colonic microbiome. This therefore highlights the need for further standardised collaborative studies, and the value of systematic reviews to provide context for further advancements into a field becoming increasingly relevant to modern day clinical practice.
List of abbreviations
- CA
-
Cancer
- CD
-
Crohn’s Disease
- CENTRAL
-
Cochrane Central Register of Controlled Trials
- CRC
-
Colorectal Cancer
- DC
-
Diversion Colitis
- DGGE
-
Denaturing Gradient Gel Electrophoresis
- FMT
-
Faecal Microbial Transplant
- FC
-
Flow Cytometry
- GI/GIT
-
Gastrointestinal/Gastrointestinal Tract
- IBD
-
Inflammatory Bowel Disease
- IHC
-
Immunohistochemistry
- IPAA
-
Ileal Pouch Anal Anastomosis
- LARS
-
Lower Anterior Resection Syndrome
- NRS
-
Non-Randomised Controlled Study
- OTU
-
Operational Taxonomic Unit
- PCR/qPCR
-
Polymerase Chain Reaction/quantitative Polymerase Chain Reaction
- PRISMA
-
Preferred Reporting Items for Systematic Reviews and Meta-analyses
- RCT
-
Randomised Controlled Trial
- SCFA (s)
-
Short-chain Fatty Acid (s)
- TPN
-
Total Parenteral Nutrition
- UC
-
Ulcerative Colitis
Acknowledgments
All authors reviewed and agree with contents of the manuscript.
Disclosure statement
The authors declare none.
Author contribution
S.W.S. contributed to data curation, formal analysis, investigation, methodology, writing original draft and review and editing. B.H. involved in data curation, investigation, formal analysis, methodology and writing original draft. H.I.O. participated in project administration, data curation, validation, visualisation, writing review and editing. S.G. contributed to resources and project administration. O.S. contributed to supervision, resources and validation. A.L. involved in conceptualisation, supervision, validation, writing review and editing. A.B. contributed to supervision and resources. D.P. contributed to supervision and resources. H.M. contributed to conceptualisation, supervision, resources, validation and writing original draft.
Funding
No funding was obtained for this study.
Data availability statement
Data from this review can be made available upon reasonable request from the corresponding author.
Open access
