Microbiome Composition in Microscopic Colitis: A Systematic Review

Believed to be a rare cause of chronic diarrhoea, microscopic colitis (MC) is a condition with rising incidence. Many prevalent risk factors and the unknown pathogenesis of MC rationalise the need for studies on microbiota composition. PubMed, Scopus, Web of Science and Embase were searched. Eight case-control studies were included. The risk of bias was assessed with the Newcastle–Ottawa Scale. Clinical details on the study population and MC were poor. The most consistent result among the studies was a decreased Akkermansia genus in faecal samples. Other results were inconsistent due to the different taxonomic levels of the outcomes. Possible changes in different taxa were observed in patients who suffered from MC compared to healthy controls. The alpha diversity compared between MC and the diarrhoea control may suggest potential similarities. The beta diversity in MC compared to healthy and diarrhoeal populations showed no significant outcomes. The microbiome composition in MC possibly differed from the healthy control, but no agreement regarding taxa was made. It might be relevant to focus on possible factors influencing the microbiome composition and its relationship with other diarrhoeal diseases.


Introduction
Chronic watery diarrhoea is the predominant demonstration of microscopic colitis (MC) [1,2]. Other associated symptoms are faecal urgency, nocturnal diarrhoea and faecal incontinence [2]. A diagnosis of MC requires a colonic biopsy, which might also reveal subtypes of MC: lymphocytic colitis (LC), collagenous colitis (CC) [1,2] or an incomplete form [2]. There is no evidence suggesting that MC might be connected with a higher risk of colorectal cancer [1,2], adenoma [2] or surgery [1]. Nevertheless, it might have a detrimental effect on patients' quality of life [1,2].
MC is sometimes believed to be a rare inflammatory bowel disease (IBD). However, the incidence of MC in Western countries has increased and is estimated at 11.4 per 100,000 person-years, with a prevalence of 119 per 100,000 persons [2], ranging from 48 to 219 per 100,000 [1]. Moreover, it occurs in 7.5% of patients who are diagnosed with chronic diarrhoea [1] and might be responsible for 12.8% of unexplained cases (high heterogeneity data) [2]. There are few risk factors associated with the development of MC, such as current and former smoking, the use of proton pump inhibitors (PPIs), nonsteroidal anti-inflammatory drugs (NSAIDs) or selective serotonin reuptake inhibitors (SSRIs) [2]. Furthermore, women are more likely to suffer from MC than men [1,2]. There are many possible factors contributing to the nonestablished pathogenesis of MC, and those that have been considered are luminal factors, extracellular matrix remodelling, involvement of the immune system and genetic predispositions [2,3]. One presumed luminal factor is microbiome composition, the alteration of which might occur in MC. Importantly, changes in the microbiome composition have been recognised in Crohn's disease (CD) and ulcerative colitis (UC) [4], highlighting the rationale behind microbiome analysis in MC.
Not only the pathogenesis but also the treatment of MC have not yet been firmly determined. Currently, the first-choice of treatment is budesonide. It was proven to be effective and safe in MC [1,2]. The response rate for budesonide treatment was 81% for collagenous colitis (CC) and 84% for lymphocytic colitis (LC) [2], while one-third of patients remained free of symptoms [1]. Treatment for nonresponders might vary. Some possibilities are available, but no strong recommendation can be made [2]. Additionally, the maintenance strategy for MC is not yet fully determined. There is evidence in favour of budesonide therapy [1,2]. One possible treatment in the future may be the administration of probiotics. However, uncertainties concerning the mechanisms of probiotics' action and efficacy still remain [5]. Nevertheless, they are considered safe agents [6], and acknowledging whether the microbiome plays a role in the pathogenesis of MC might put probiotics use into perspective.
Based on the available data, we decided to cover the aspect of microbiota composition in microscopic colitis to provide practical insight into the further guidance of patients. The microbiota composition in microscopic colitis has not yet been fully determined. A new understanding of microbiota's role in the pathogenesis of MC might open the door for personalised treatment.

Search Strategy
This review was based on a search of PubMed, Scopus, Web of Science and Embase to assess studies on the microbiome. The results of the search of the databases were acquired on 19 January 2023 for microbiome-related studies. No database filters were used.

Study Selection
The results of the conducted search are presented as a flow diagram according to the PRISMA statement ( Figure 1). All of the results were assessed by two researchers. First, the duplicates were removed with the help of Zotero software. Second, the titles were screened to choose articles possibly related to the topic. Studies in English or German were included. The records regarding paediatric populations and the assessment of the microbiome by cultivation were excluded. Any doubt regarding the article's title was an indication to include it in the next step. Finally, the abstracts were screened, and the articles were chosen for a full-text assessment. During this phase, case reports and case series were excluded. For occurring doubts concerning the inclusion of an article, the other authors were consulted.

OR bacteriobiome OR bacteriome OR bacterial microbiome OR microflora OR intestine microflora OR feces microflora Search: #1 AND #2
The results of the search of the databases were acquired on 19 January 2023 for microbiome-related studies. No database filters were used.

Study Selection
The results of the conducted search are presented as a flow diagram according to the PRISMA statement ( Figure 1). All of the results were assessed by two researchers. First, the duplicates were removed with the help of Zotero software. Second, the titles were screened to choose articles possibly related to the topic. Studies in English or German were included. The records regarding paediatric populations and the assessment of the microbiome by cultivation were excluded. Any doubt regarding the article's title was an indication to include it in the next step. Finally, the abstracts were screened, and the articles were chosen for a full-text assessment. During this phase, case reports and case series were excluded. For occurring doubts concerning the inclusion of an article, the other authors were consulted.

Data Extraction and Analysis
All articles that were assessed as eligible for this review were screened for the study's characteristics, population description, features of MC and associated diarrhoea, microbiome composition based on sequencing and the alpha and beta diversity indices. The derived details were double-checked for mistakes. Any doubts were dispelled by the coauthors or described in the study. Whenever needed, the authors of the included articles were contacted by email in order to ask about details.
Due to the heterogeneity of the data regarding the microbiome, a meta-analysis was not performed. Therefore, only a narrative characterisation of the outcomes was included in the present study.

Data Extraction and Analysis
All articles that were assessed as eligible for this review were screened for the study's characteristics, population description, features of MC and associated diarrhoea, microbiome composition based on sequencing and the alpha and beta diversity indices. The derived details were double-checked for mistakes. Any doubts were dispelled by the co-authors or described in the study. Whenever needed, the authors of the included articles were contacted by email in order to ask about details.
Due to the heterogeneity of the data regarding the microbiome, a meta-analysis was not performed. Therefore, only a narrative characterisation of the outcomes was included in the present study.

Risk of Bias
The quality of the studies was assessed independently for the microbiome using the Newcastle-Ottawa Quality Assessment Scale for case-control studies (CCS). A full score using this scale is 10 points. A study with ≥6 points was regarded as a good-quality study.

Description of Study Design and Population
The performed selection of studies resulted in the inclusion of eight articles regarding microbiome composition [7][8][9][10][11][12][13][14] (Table 1). Every included study was conducted in Western countries and designed as a CCS. First, an analysis and description of the study population were performed. The study population examined for the microbiome composition was n = 183. Most of the studies assessed the microbiome composition without a targeted intervention and compared it to the control group [7,10,11,13,14]. Different control groups were applied in separate studies. Healthy controls were the most prevalent and consisted of n = 121. Other control groups regarded diseases associated with diarrhoea, including bile acid diarrhoea (n = 16) [9], functional diarrhoea (n = 31) [9,14], Crohn's disease (n = 53) [10,11] and ulcerative colitis (n = 70) [10,11]. In [7], the control group was described as patients reporting loose stools (n = 153). The overall description of the clinical details of MC was poor ( Table 2). Each study was conducted in an adult population only. Women were more prevalent in seven out of eight studies, except the study by Millien et al. [8]. Only four studies reported body mass index (BMI), indicating a wide range of patient inclusion (underweight, overweight and obese) [7,9,11,14]. Autoimmune disorders and budesonide treatment were reported in one study only [11]. Additionally, each patient in the study by Krogsgaard et al. [12] was treated with budesonide after the first microbiome assessment. Few studies reported the prevalence of risk factors among the participants. Each of them might act as an agent, which influences the microbiome composition. Current and former smoking were indicated in four studies [8,[12][13][14]. The only reported drugs taken were proton pump inhibitors (PPIs) [8,12] and nonsteroidal anti-inflammatory drugs (NSAIDs) [8,9]. The elimination of the use of antibiotics before inclusion was mandatory in six studies: 3 months [7-9,11,12] and 1 month [10]. Probiotics were eliminated for 3 months [9] or 1 month [12].

Details about Microscopic Colitis
The following clinical details about the MC course were described in the included studies (Table 3.). The subtype was defined in four out of eight studies. One study included patients diagnosed with collagenous colitis only [11]. Currently, there are no differences in the clinical management for any of the subtypes [1]. However, some diversity in the evidence for LC and CC occurred [2]. Three studies verified the diagnosis of MC through biopsy before inclusion [7][8][9]. Regarding the clinical activity of MC, it should be assessed using the Hjorstwang criteria [15]. However, they were applied only by Carstens et al. [11] and Krogsgaard et al. [12]. The duration of diarrhoea was assessed in one study, with a median of 20 weeks [9]. In the study by Milien et al. [8], 40% of patients had diarrhoea for less than 12 months. Daily bowel movements were described in two studies [8,9].

Microbiome Composition Analysis
The results of the microbiome assessment differed among the studies (Table 4). Biopsy [7,8] or faecal samples [9][10][11][12][13][14] were obtained. The method of the microbiome analysis was based on sequencing technology. Most of the studies applying sequencing methods relayed 16s RNA [7][8][9][10][11] or 16s DNA [12]. However, two studies were based on metagenomic sequencing [13,14]. The outcomes of the alpha diversity assessment for MC and healthy controls showed no differences in three studies [8,9,11]. However, in [10,12] the alpha diversity was significantly lower in MC than in the healthy control. One study [12] proved that the alpha diversity was higher in MC after treatment with budesonide than at the beginning. Differences were not detected for the beta diversity [8]. The microbial dysbiosis index was noticed to be higher in active MC than in the remission phase [14]. Additionally, there were no differences observed between LC and CC in the alpha diversity [9,10], beta diversity [9], microbial dysbiosis index [9] and bacterial abundance [12]. Second, the differences between MC and other diarrhoea controls (DCs) were checked. There were no differences in the alpha diversity between MC and CD or UC [10] or regarding the functional DC and bile acid DC [9]. In one study, the alpha diversity for MC, functional DC and bile acid DC together was lower than for HC [9]. The microbial dysbiosis index was also calculated for the DC and was proved to be higher in the active phase of MC than in the functional DC [14] and higher in MC than in the functional DC together with the bile acid DC [9]. These results indicate that there might be some differences between MC and HC. Moreover, there is the possibility that there were some similarities between MC and the DC, or another factor, in common that influenced the results.
The outcomes of the detailed microbiome composition were presented as increased or decreased taxa in MC compared to the healthy control. Most of the studies used a level of significance of <0.05 [9,[11][12][13]. Except for [13], all studies reported adjusting results with a correction for multiple testing. All studies used different taxonomic levels in their reports. Therefore, the reported taxa were analysed for their taxonomic levels to assess whether patterns among the described taxa occurred. For this purpose, the National Center for Biotechnology Information (NCBI) Taxonomy's browser [16], List of Prokaryotic Names with Standing in Nomenclature (LPSN) [17] and the NCBI's Lifemap were used [18]. The details of the analyses were reported as follows: Concerning increased taxa, two studies showed no differences in the faecal samples [11,13]. The only similarities among the studies were an observed increase in the Veillonella genus [10] and the identification of Veillonella unclassified and V. parvula species in faecal samples [14].
Concerning the decreased bacterial taxa in MC compared to healthy controls, potential patterns were noticed. The most consistent among the studies was a decrease in the Akkermansia genus from the Verrucomicrobia phylum in the faecal samples [10][11][12][13]. In addition, the whole Verrucomicrobia phylum was described as decreased in [13]. A reduction in the Coriobacteriaceae family in both biopsy [8] and faecal samples [10] was described. Moreover, in the Coriobacteriaceae family, lower levels of Coriobacteriaceae unclassified [11] and Collinsella genus [10,11] were observed in the faecal samples. A decrease in the Ruminococcaceae [10] family; Ruminococcaceae NK4A214, UCG-002, UCG-005 and UCG-010 groups [12]; Ruminococcaceae unclassified [11]; and Ruminococcus 1 and 2 genus [12] in the faecal samples was described. Additionally, an increase in the Ruminococcaceae family and Ruminococcus 2 genus was observed between baseline and after treatment with budesonide, which might indicate shifts in this taxa [12]. Furthermore, in the decreased Rikenellaceae family [10], a reduction in the Alistipes genus [11] and Alistipes putredinis species [14] was observed. Finally, Clostridiales unclassified [9,11], Ruminiclostridium genus [8], Clostridium genus [12] and Clostridiales unclassified [10] were decreased in the faecal samples. Whether there were actual changes for the mentioned taxa is uncertain.
Few studies also compared outcomes between MC and DC. However, no consistent results occurred. Similarities occurred with regards to the decreased OTUs from the Ruminococcaceae family in CD and UC [11] and an increased abundance of Alistipes putredinis in the functional DC [14]. Some taxa were described as increased or decreased in MC compared to the unspecified diarrhoea control [7]. One study reported no significant differences between the relative abundances in MC and the functional assessed together with bile acid DC [9].
In conclusion, there were observable common patterns for the increased and decreased taxa. More results are described in the "Additional Outcomes" column in Table 4.    Abundance of Alistipes putredinis increased in functional DC (q = 0.043). Microbial dysbiosis index was higher in active MC than in functional DC.
No differences at the species level for active and remission MC. Alpha diversity was lower in active than in remission MC (q = 0.031). No other differences. Increase in global PTR (q = 0.033) and Alistipes finegoldii-specific PTR (q = 0.065) were observed in active MC compared to HC. 1 Microbiome of MC was compared to microbiome in HC as four different models adjusted for the population's features (1: education, proton pump inhibitor use and batch; 2: 1 + sex and antibiotics use; 3: 2 + age; 4: 3 + BMI). Only the outcomes that were in common for models 1, 2 and 3 or 1, 2, 3 and 4 are included in Table 4. 2 The level of Akkermansia in smokers was assessed in larger groups than those included in the study; an additional 5 female with MC and 7 HC were included for the PCR analysis. 3 Log of (total abundance in organisms increased) over (total abundance of organisms decreased). FDR, false discovery rate; MC, microscopic colitis patients; HC, healthy control; DC, diarrhoea control; CC, collagenous colitis; LC, lymphocytic colitis; FD, functional diarrhoea; CD, Crohn's disease; UC, ulcerative colitis; NSAIDs, nonsteroidal anti-inflammatory drugs; PPI, proton pump inhibitor.

Discussion
The analysis of the available data on the microbiome composition in MC in comparison to healthy controls poses the questions of whether there were actual differences among them and whether possible similarities existed between MC and other diarrhoeal diseases.
On a large scale, the changes in the alpha and beta diversities describe shifts in the microbiome composition. The findings differed among studies, showing either a lower alpha diversity compared to the healthy control [7,10,12] or no significant differences [8,9,11]. There are many possible variables that might influence the outcome. It was observed that the Shannon diversity index was significantly correlated with food and drink intake, physical activity, medication used, symptoms and blood clinical markers [19]. Therefore, the poor descriptions of the clinical details in the included studies might be a factor altering the conclusions. Nevertheless, the details regarding BMI indicated a wide diversity in patients. Therefore, the observed potential negative correlation between the Shannon index and both weight and BMI [19] might reveal a possible influencing factor on our findings. On the contrary, a recent meta-analysis showed no differences in the Shannon index in obese patients [20]. The relationship between obesity and alpha diversity might depend on the metabolic status of the individual, with a lower diversity among the unhealthy [21]. Therefore, the impact of BMI on the microbiome might not be straightforward. Not only BMI but also the appearance of diarrhoea in a week, number of bowel movements per week and abdominal pain were negatively correlated with alpha diversity status [19]. The stool's form might be considered a possible factor influencing the assessment of the microbiome. Regarding CD and UC, a lower observed richness was detected in a recent meta-analysis [22]. Interestingly, one included study showed that the alpha diversity was similar for MC and CD/UC [10]. In another study, no differences between MC and functional DC or bile and acid DC were observed [9]. These findings support the question regarding similarities between MC and other diarrhoeal diseases. As for the beta diversity, the included studies showed no differences among the faecal samples [9] or biopsy samples [8]. In comparison, a lower beta diversity was observed in UC and CD compared to the non-IBD control, with a significant impact due to the samples' type: faecal or biopsy [22].
Currently, the impact of the sample type on the microbiome assessment outcomes is being recognised as an important variable in studies. This might indicate the need to obtain biopsy samples for an analysis of microbiome composition instead of faecal samples due to the better representativeness of the colon microbiome [23]. On the contrary, emerging evidence shows that bowel preparation for a colonoscopy might not be without an influence on the outcomes of a microbiome assessment from biopsy [24] and faecal samples [24,25]. Moreover, the colonoscopy required for biopsy collection is an invasive procedure that is inadequate for healthy controls, with a small number of possible sampling sites and occurring concerns regarding contamination [26]. Here, included were two studies that used biopsy collection for a microbiome composition assessment [7,8]. Importantly, one of the studies used paraffin-embedded colonic tissue samples, which might be altered by the procedure [8]. It was difficult to conclude whether there were differences among the studies due to the small number of studies that were based on biopsy samples. Additionally, none of the studies described the time from the last diagnostic colonoscopy, except in [9]. Batista et al. [9] collected faecal samples three days before the diagnostic colonoscopy and 30 days after, thus limiting the impact of colonic cleansing. However, the faecal microbiome 30 days after a colonoscopy showed no changes in its composition [9]. Both the biopsy and faecal sampling might have influenced the outcomes of the included studies The determination of the microbiome composition at a taxonomic level was beyond the scope of this review. However, possible patterns were recognised in the included studies that might be worth exploring. The most consistent result regarding taxa concerned a decrease in the Akkermansia genus from the Verrucomicrobia phylum in faecal samples [14,[16][17][18]. Importantly, the presence of Akkermansia was observed in the healthy population [27]. Currently, the correlation between Akkermansia and obesity is being thoroughly examined.
A lower level of Akkermansia was observed with obesity and associated with age and relative abundance. However, the correlation between Akkermansia and being overweight was insignificant [28]. Here, a description of BMI was present in only one of the studies, where BMI ranged from 17 to 35 [11]. Therefore, patients with obesity were included in the study population. Interestingly, in the study by Carstens et al. [11], the impact of BMI was proved to be significant between participants with BMI < 30 and >30, but no differences occurred between the study and control populations concerning BMI. As no other study described BMI, it should be taken into consideration as a variable that might influence the level of Akkermansia. In comparison with other gastrointestinal diseases, a decrease in Akkermansia was also observed in a meta-analysis of UC [22]. It might be crucial that the A. muciniphila species belongs to mucin-degrading bacteria [29,30]. Therefore, it is involved in the maintenance of the mucus layer and might play a role in gut homeostasis. Whether Akkermansia and its species have a potential role in colitis is the subject of ongoing, robust studies.
The Veillonella genus is known to be a part of a healthy oral microbiome [31]. Moreover, Veillonella was also associated with a healthy microbiome detected in faecal samples [32], but no conclusion on its oral or colon origin could be made due to the impossibility of their detection at the species level [27]. In general, stool microbiota differs from upper tract microbiota [27]. The link between the oral and gut microbiomes is yet to be determined [27,33]. There is a hypothesis that considers an influence through bacterial or immune pathways [33]. As PPI administration is a risk factor for MC, it might be worth noticing that PPI was connected with a higher prevalence of oral cavity bacteria in the gastrointestinal microbiome [34]. Even though Veillonella was recognised in a healthy population, a meta-analysis described an increased abundance of Veillonella in CD and UC [22]. Therefore, it might be interesting that Veillonella and Veillonella unclassified were reported to be increased in faecal samples in [10] and [14], respectively. At the species level V. parvula was detected to be increased in faecal samples [14]. Recently, it was also described as elevated in faecal samples from CD [35] and associated with the inflammatory potential of diet in CD [36]. Whether there is an actual connection between Veillonella and gut inflammation might be a perspective for future studies.
Microbiome changes might imply the relevance of therapy that focuses on the gut bacteria through the administration of probiotics. Currently, the administration of probiotics in CD and UC is controversial. The American Gastroenterology Association (AGA) only supports the use of probiotics in clinical trials [37]. In comparison, the European Crohn's and Colitis Organisation (ECCO), in a review on complementary medicine and psychotherapy in IBD, concluded that there is not enough evidence for the use of probiotics in CD, but Escherichia coli Nissle 1917 or multi-strain probiotics containing a combination of lactic acid bacteria, Streptococcus and Bifidobacteria might be considered as an alternative for the induction and maintenance of remission in UC [38]. Large RDBCTs with follow-up are required to determine the influence of probiotics in MC. Interestingly, engineered probiotics have recently been investigated in mouse models for the treatment of IBD [39]. These probiotics might be the future for treatments related to the microbiome.
There were several limitations to this study. First, no synthesis in a meta-analysis format was possible. Second, the descriptions of the study population and clinical features were poor. Therefore, the detection of some important factors with a potential influence on the outcomes was difficult. Furthermore, various exclusion and inclusion criteria were also used for the study population. Therefore, this might have interfered with the comparison of the studies. Finally, the descriptions of the taxonomy levels varied among the studies, which sometimes made it difficult to compare the results. Therefore, no specific pathogenic species were determined. In conclusion, our findings might be considered as an overview and guide for future studies with a strong focus on microbiome composition in patients suffering from MC.

Conclusions and Future Directions
The microbiome composition is potentially altered in MC; however, no firm agreement on the composition or taxa related to the pathogenesis or course of MC can be made. Future studies could focus on factors that influence the composition of the microbiome. It might be relevant to establish the relationship between the risk factors of MC and the microbiome, which might provide better insight into the pathogenesis of MC. Further research regarding the correlation between MC and other gastrointestinal diseases might also be important for a better understanding of MC. Moreover, it might be important to determine the pathogenic species for MC. Consequently, such results might become a guide for the treatment and guidance of patient.