If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Internal Medicine, University of Manitoba, University of Manitoba IBD Clinical and Research Centre, Winnipeg, MB, CanadaNational Microbiology Laboratory, Public Health Agency of Canada, Winnipeg, MB, CanadaDepartment of Laboratory Medicine & Pathobiology, University of Toronto, ON, Canada
National Microbiology Laboratory, Public Health Agency of Canada, Winnipeg, MB, CanadaDepartment of Medical Microbiology and Infectious Diseases, University of Manitoba, Winnipeg, MB, Canada
National Microbiology Laboratory, Public Health Agency of Canada, Winnipeg, MB, CanadaDepartment of Medical Microbiology and Infectious Diseases, University of Manitoba, Winnipeg, MB, Canada
Correspondence to: Djavad Mowafaghian Centre for Brain Health, Faculty of Medicine (Neurology), University of British Columbia, Room S126, UBC Hospital, 2211 Wesbrook Mall, Vancouver, BC V6T 2B5, Canada.
No major differences in the overall gut microbiota composition in MS cases vs. controls.
•
Abundance of six taxa differed in MS cases vs. controls, similarly observed across 2 ≥ studies.
•
Studies were generally too modest in size to adequately assess potential effect modifiers.
Abstract
Background
To systematically review and synthesize the literature on the multiple sclerosis (MS) gut microbiota composition as compared to persons without MS.
Methods
We systematically searched MEDLINE, EMBASE, and Web of Science databases for relevant published articles (2008–2018).
Results
Of 415 articles identified ten fulfilled criteria. All studies used a case-control design, six sourced participants from the US, two Germany, one Italy, and one Japan. Nine focused exclusively on adults and one on children, totaling 286 MS and 296 control participants. Over 90% of cases had relapsing-remitting MS; disease duration ranged from 10.6 ± 6.5 months to 15.3 ± 8.6 years (mean±SD). Nine studies examined stool and one evaluated duodenal mucosa. Diverse platforms were used to quantify microbes: Illumina MiSeq, Roche 454, microarray, and fluorescence in situ hybridization. None of eight studies reported a significant alpha-diversity differences between cases and controls. Two of seven studies reported a difference in beta-diversity (P ≤ 0.002). At the taxa-level, ≥2 studies observed: lower relative abundance of Prevotella, Faecalibacterium prausnitzii, Bacteroides coprophilus, Bacteroides fragilis, and higher Methanobrevibacter and Akkermansia muciniphila in MS cases versus controls. Exposure to an immunomodulatory drug (IMD), relative to no exposure, was associated with individual taxonomic differences in three of three studies.
Conclusion
Gut microbiota diversity did not differ between MS cases and controls in the majority of studies. However, taxonomic differences were found, with consistent patterns emerging across studies. Longitudinal studies are warranted to elucidate the relationship between IMD exposure and differences in the gut microbiota composition.
). An important initial step to understanding the relationship between the gut microbiota and MS is to survey the gut microbial community. Recent studies on MS have focused on surveying the gut bacterial (and archaeal) communities. A common goal has been to investigate whether there are differences in the MS gut microbiota composition between MS cases and controls, measured as the overall microbiota composition (diversity) and relative abundance of individual resident microbes. Some studies also assessed potential effect modifiers (or confounders), such as exposure to an immunomodulatory drug (IMD) used to treat MS.
We conducted a systematic review in order to comprehensively collate the body of evidence surrounding the relationship between the gut microbiota and MS. Our objective was to include published articles in which the gut microbiota profiles had been compared between individuals with and without MS.
2. Methods
Our systematic review was designed to address the following specific questions: (1) Does the gut microbiota composition differ between MS cases and controls (participants without MS), as assessed by (a) diversity metrics and/or by (b) taxa-level relative abundances? (2) What are the main effect modifiers (confounders) identified to date in studies evaluating the gut microbiota in MS?
This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines (
). The systematic review protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO) database (registration no. CRD42018089173) and the protocol is in accordance with the PRISMA-P (2015) guidelines (
Original research articles assessing the potential gut microbiota differences in diversity or taxonomic relative abundance between MS cases and controls (individuals without MS) were eligible for inclusion. We included any of the following study designs: cohort, cross-sectional, case-control, and comparative cohort studies. Other study designs were excluded, such as intervention studies (unless pre-intervention samples were available for cases/controls) and studies without a control group (e.g., case series). No restrictions were imposed with respect to the age of study participants, geographical location or setting (e.g., community or hospital), or disease course(s) under study (e.g., relapsing-onset, primary progressive).
2.2 Literature search strategy
We performed a systematic literature search of MEDLINE, EMBASE, and Web of Science databases for original research articles published in English between January 1st, 2008 and February 8th, 2018, which we subsequently updated to June 19th, 2018. At the time of publication, a final update to August 24th, 2019 was included and reported as an addendum. The search strategies for each database are shown in Fig. 1A. We did not include conference abstracts, unpublished work or the grey literature (e.g., presentations, posters, websites or dissertations).
Fig. 1Literature search strategies and PRISMA flow diagram.
A. Systematic literature search strategies of MEDLINE, EMBASE, and Web of Science databases for original research articles. B. Selection of articles for inclusion in the systematic review of the multiple sclerosis microbiota (2008–18).
2.3 Eligibility assessment, data extraction, and quality assessment
The literature search results were uploaded to Mendeley for screening. Titles and abstracts were screened based on the study inclusion and exclusion criteria by two independent reviewers (AM and JF). Any disagreement was resolved between the reviewers. The full texts of all screened-in abstracts were then retrieved and assessed for eligibility and relevant information retrieved were extracted by one reviewer (AM), as outlined in Appendix 1. Two independent reviewers (AM and JF) assessed the risk of bias in individuals studies specific to our current systematic review-related questions using the US's National Institutes of Health (NIH) tool for Quality Assessment of Case-Control Studies (
). No study was excluded based on the risk of bias (rated as good, fair, and poor), in part because a study could be assigned as ‘poor’, inferring a high risk of bias, if the microbiota quantification platform was not valid to address our study question(s).
3. Results
3.1 Study and participant characteristics
Of 415 articles identified (based on titles and abstracts), ten fulfilled criteria for inclusion (
Dysbiosis in the gut microbiota of patients with multiple sclerosis, with a striking depletion of species belonging to clostridia XIVa and IV clusters.
). All ten studies used a case-control design for a total of 286 MS cases and 296 controls. Nine studies enrolled adults, of which one focused solely on twins discordant for MS, and one enrolled children. Nine studies collected stool samples from participants while one collected duodenal mucosal samples. Characteristics of the included studies are summarized in Table 1. Study quality was ‘fair’ for nine articles and ‘poor’ for one. The study identified as poor used an obsolete method (fluorescence in situ hybridization; FISH) for characterizing the gut microbiota, thus making it a high risk of bias in the context of our systematic review-specific questions (
were missing clinical data for 8 subjects; they did not report which clinical data was missing. Continuous data are expressed as mean ± SD. NA, not available; MS, multiple sclerosis; BMI, body mass index; NIH, National Institutes of Health.
Across the nine studies which reported participant demographics, females predominated. Overall the number (%) of females/males were: 182/104 (64%/36%) for cases and 181/111 (61%/38%) for controls (sex was missing for four controls in one study (
)). The average ages of cases and controls ranged from 12–54 and 13–54 years, respectively across studies (age was missing for nine controls and four cases in one study (
)). Across the four studies that reported race and/or ethnicity, the majority of participants were Caucasian, totaling 87/101 (86%) of MS cases and 75/84 (89%) of controls. Six studies recruited participants from the United States, two from Germany, one from Italy, and one from Japan. The most commonly reported lifestyle factors were diet (five studies) and BMI (five studies), both diet and BMI were available in three studies (Table 1). Diet was collated via a validated food frequency questionnaire in two US-based studies; three did not specify. Both diet metrics and BMI (based on descriptive group means) were similar when cases were compared to controls, except for one study, which, for MS cases, observed a higher intake of specific dietary elements (i.e., carotenoids) (
The majority (267/286; 93%) of MS cases had a relapsing-remitting disease course at the time of sample collection. Five studies used the McDonald 2010 criteria for MS diagnosis, one used the Poser criteria and four did not specify (
(range 0–5.5) 0.0 to < 2.0, n=8; 2.0 to < 3.0, n=4; 3.0 to < 6.0, n=7
RR/NEDA 9 (47%), RR/EDA 10 (53%)
19 (100%)
7 (37%)
9 (47%)
0
0
3 (16%)
0ᶜ
Swidsinski, 2017
NA
NA
NA
NA
RR 10 (100%)
NA
NA
NA
NA
NA
NA
NA
NA
Berer, 2017
28.0 ± 9.0
13.2 ± 9.6
McDonald
NA
CIS 3 (9%), RR 22 (65%), SP 7 (21%), PP 2 (6%)
19 (56%)ᶜ
13 (38%)
1 (3%)
4 (12%)
0
0
1
0ᶜ
PP 2 (6%)
Cekanaviciute, 2017
NA
NA
NA
NA
RR 71 (100%)
0 (0%)ᶜ
0
0
0
0
0
0
0ᶜ
Total
118 (43%)
64 (23%)
35 (13%)
10 (4%)
5 (2%)
3 (1%)
1
11 (5%)
Unless otherwise stated, characteristics shown relate to the time of microbiota sample collection. The frequency of exposures to an immunomodulatory drug, immunosuppressant and systemic corticosteroids at the time of microbiota sample collection were typically within 2 or 3 months prior to sample collection. One study did not report treatment status. Systemic corticosteroids are not included in the number of MS cases exposed to IMD/IMS (column 7). All percentages are rounded to the nearest whole numbers. Continuous data are expressed as mean ± SD. NA, not available; MS, multiple sclerosis; RR, relapse-remitting multiple sclerosis; SP, secondary progressive multiple sclerosis; PP, primary progressive multiple sclerosis; CIS, clinically isolated syndrome; RR-EDA, evidence of disease activity in patients with relapse-remitting multiple sclerosis; NA, not available; RR-NEDA, no evidence of disease activity in patients with relapse-remitting multiple sclerosis; F, female; M, male; EDSS, Expanded Disability Status Scale. IMD, Immunomodulatory drug; IMS, immunosuppressant; ᵃExposed within 1 month; ᵇExposed within 2 months; ᶜExposed within 3 months.
Medication exposure status was reported in nine of ten studies; all focused on IMDs or immunosuppressants (IMS) used to treat MS. Overall, 118 (43%) of MS cases (n = 276) were exposed to an IMD or IMS near the time of sample collection; 64 (23%) to β‐interferon, 35 (13%) glatiramer acetate, 18 (7%) other IMDs, and 1 (<1%) IMS (azathioprine). IMD or IMS exposure was typically defined as exposed in the two or three months prior to the time of microbiota sample collection. The pediatric study included IMD naïve MS cases, explicitly defined as never exposed to an IMD (
). Of the seven studies that reported use of systemic corticosteroids, 11/229 (5%) of MS cases were exposed to systemic corticosteroids near the time of sample collection (Table 2).
3.2 Sample handling and sequencing procedures
Eight of the ten studies mentioned at least some aspects of the methods related to sample collection. A US-based group sampled participant's microbiota using culture swabs, another US-based group collected stool directly into a dry container, and the Italy-based group collected mucosal biopsies into a culturing solution containing antibiotics (
). Four studies shipped stool samples overnight (shipped with ice with the exception of one study) while four groups stored or processed the sample the same day of the collection (
Dysbiosis in the gut microbiota of patients with multiple sclerosis, with a striking depletion of species belonging to clostridia XIVa and IV clusters.
Seven studies sequenced various regions of the 16S rRNA gene using a Next-Generation Sequencing (NGS) platform (either Illumina MiSeq or Roche 454). One study used both Roche 454 and Illumina MiSeq (Table 3) (
). The 16S rRNA variable-region gene sequenced varied between studies. The method of generating ‘Operational Taxonomic Units’ (OTUs) also varied. OTUs are clusters of similar sequences—typically > 97%—that are assumed to represent a single taxon; for 16S rRNA, OTUs typically have genus-to-species level of taxonomic specificity. The different methods of generating OTUs among the studies that used an NGS platform include: de novo OTU clustering (four studies) and closed reference OTUs (three studies). Despite these differences, all studies that used an NGS platform clustered the 16S rRNA sequences into OTUs based on a 97% similarity threshold (Table 3).
Table 3Technical and computational methods used to process and quantify the microbiota.
First Author, Date of Publication
Microbiota Quantifying Instrument, 16S rRNA Region
Method Generating OTUs
Normalization of Sample Sequence Depth
Cantarel, 2015
DNA microarray (Affymetrix PhyloChip)
The array contains representative sequences from OTUs, which were clustered at 95% similarity
NA
Miyake, 2015
Roche 454, V1-V2
Individual taxa analysis: Closed reference OTUs at 97% similarity. Diversity analysis: De novo OTU clustering at 96% similarity
Rarefied to 3000 sequence per sample
Chen, 2016
Illumina MiSeq,V3-V5
De novo OTU clustering at 97% similarity using IM-TORNADO
Rarefied to 10,000 sequence per sample
Cree, 2016
DNA microarray (Affymetrix PhyloChip)
NA
NA
Tremlett, 2016
Illumina MiSeq, V4
De novo OTU clustering at 97% similarity using QIIME
Rarefied to 201,546 sequence per sample
Jangi, 2016
Roche 454, V3-V5;Illumina MiSeq, V4
De novo OTU clustering at 97% similarity using Mothur
NA
Cosorich, 2017
Roche 454, V3-V5
De novo OTU clustering generaed using QIIME
At least 3000 sequence per sample
Swidsinski, 2017
Fluorescence in situhybridization (FISH)
NA
NA
Berer, 2017
Roche 454, V3–V5
Closed reference OTUs with 97% similarity using QIIME
Diversity analysis: Rarefied to 10,975 sequences per sample.Individual taxa analysis: variance-stabilizing transformation
Cekanaviciute, 2017
Illumina MiSeq, V4
Closed reference OTUs with 97% similarity using QIIME
Diversity analysis: Rarefied to 10,000 sequence per sampleIndividual taxa analysis: variance-stabilizing transformation
Closed-reference OTU assignment assigns query sequences to OTUs generated from an external reference database, whereas sequences in de novo OTU clustering are clustered against one another. The two common methods for normalizing read counts were rarefying and scaling. Rarefying refers to randomly discarding reads in each sample until each sample has an equal number of sequences. Scaling often includes transforming, e.g., variance stabilizing transformation. OTU, operational taxonomic units; NA, not available; V, variable region of 16S rRNA gene; IM-TORNADO, Illinois Mayo Taxon Organization from RNA Dataset Operations; Qiime, Quantitative Insights Into Microbial Ecology.
The twin cohort study also sequenced the gut metagenome (the collective genomes of all the gut microbes in a stool sample), although authors reported that none of the results reached significance, and no data was provided (
Eight of ten studies assessed the gut microbiota diversity (alpha- or beta-diversity), Table 4. None of the alpha-diversity metrics, e.g., the number of different species within a sample, differed significantly between cases and controls. Out of seven studies that calculated beta-diversity (the number of microbial species that are not the same in two different environments), two studies reported a difference in a beta-diversity metric, between cases and controls (Table 4) (
Dysbiosis in the gut microbiota of patients with multiple sclerosis, with a striking depletion of species belonging to clostridia XIVa and IV clusters.
). No studies reported whether potential confounders, such as IMD exposure, were relevant to the significant alpha- or beta-diversity findings.
Table 4Gut Microbiota Diversity: MS cases versus controls.
First Author, Date of Publication
Diversity Metric
MS vs Controls (Main Analyses)
IMD-Related Subgroup Analyses
Statistical Test Used
Main Findings (R and P-values)
IMD Exposed MS vs IMD Unexposed MS
IMD Unexposed MS vs Control
Cantarel, 2015
β-diversity (weighted UniFrac)
PERMANOVA
P = 0.74
P = 0.66
NA
Miyake, 2015
α-diversity (richness: Choa1)
Welch's test
P > 0.05
NA
NA
α-diversity (diversity: Shannon index)
Welch's test
P > 0.05
NA
NA
β-diversity (weighted UniFrac)
ANOSIM
R = 0.24, P ≤ 0.0009
NA
NA
β-diversity (unweighted UniFrac)
ANOSIM
R = 0.21, P ≤ 0.002
NA
NA
Chen, 2016
α-diversity (richness: observed OTUs)
NA
P = 0.73
NA
NA
α-diversity (diversity: Shannon index)
NA
P > 0.05
NA
NA
β-diversity (Bray-Curtis)
PERMANOVA
P<0.001
NA
NA
Cree, 2016
NA
NA
NA
NA
NA
Tremlett, 2016
α-diversity (evenness)
Mann-Whitney
P > 0.2
NA
P > 0.05
α-diversity (richness)
Mann-Whitney
P > 0.2
NA
P > 0.05
α-diversity (Faith's phylogenic diversity)
Mann-Whitney
P > 0.2
NA
P > 0.05
β-diversity (Canberra)
PERMANOVA
P > 0.05
P=0.016
NA
Jangi, 2016
α-diversity (diversity: Shannon index)
Wilcoxon rank-sum test
P > 0.05
NA
NA
β-diversity (weighted UniFrac)
AMOVA
P > 0.05
NA
NA
β-diversity (unweighted UniFrac)
AMOVA
P > 0.05
NA
NA
β-diversity (Bray-curtis)
AMOVA
P > 0.05
NA
NA
Cosorich, 2017
α-diversity (richness: observed OTUs)
Student's t-test
P > 0.05
NA
NA
Swidsinski, 2017
NA
NA
NA
NA
NA
Berer, 2017
α-diversity (Faith's phylogenetic diversity)
NA
P > 0.05
NA
NA
β-diversity (weighted UniFrac)
NA
P > 0.05
NA
NA
Cekanaviciute, 2017
α-diversity (richness: Choa1)
NA
NA
NA
P > 0.05
β-diversity (unweighted UniFrac)
NA
NA
NA
P > 0.05
Diversity tests that were statistically significant (P < 0.05) are in bold. Blank cells indicate a diversity test result was not reported. Not all diversity metrics were reported for every alpha-diversity measure. IMD Exposed: MS cases exposed to an immunomodulatory drug within 3 months of sample collection. IMD, Immunomodulatory drug; ANOSIM, The ANalysis Of SIMilarity; PERMANOVA, PERmutational Multivariate ANalysis Of VAriance; AMOVA, Analysis of MOlecular Variance; NA, not available.
). Alpha-diversity did not differ between IMD exposed and unexposed MS cases in either study, although beta-diversity did differ in the pediatric MS study (Table 4) (
Taxa-level differences between cases and controls were reported as follows: phylum level (seven studies), genus (seven studies), and species/OTU level (eight studies). In total, two genera and four species (none of which reached significance at the phylum level) similarly differed in their relative abundances between MS cases compared to controls in a consistent direction for two or more studies (Table 5). Specifically, at the genus level, lower relative abundances of Prevotella were observed in three studies and a higher relative abundance of the Archaea Methanobrevibacter was observed in two studies. At the species level, there was a higher relative abundance of Akkermansia (muciniphila) and Faecalibacterium prausnitzii, which was observed in four studies each. Also, a lower relative abundance of Bacteroides coprophilus was reported in three studies and a lower Bacteroides fragilis abundance was reported in two studies (Table 5). Different statistical tests were used to assess differences in OTU abundance, including: models based on the negative binomial distribution (four studies), t-tests (Welch or Student, three studies), Wilcoxon rank sum test (two studies) and ANOVA (one study, Table 6).
Table 5Key findings from taxa-level relative abundances: MS cases versus controls (all MS cases versus controls are initially shown, regardless of immunomodulatory drug (IMD) exposure. In addition, findings from IMD unexposed MS cases relative to controls are also depicted).
Taxa listed here are differentially significant across 2 or more studies. Red and green cells indicate a lower and higher relative abundance, respectively. Empty cells indicates that the respected study did not report the differential abundance of the respected taxa between the two groups. ‘MS vs Control’ refers to all of the MS participants included in the study, regardless of IMD exposure, compared to the control participants. ‘IMD unexposed MS vs Control’ refers to the MS cases unexposed to an IMD at the time of microbiota sample collection, compared to the controls. Genus Akkermansiais represented as Akkermansia muciniphila. The 4 studies that compared taxa-level findings between IMD unexposed MS cases vs. controls are (1st author, num. MS cases vs. controls): Tremlett, n = 9 vs. n = 17; Jangi, n = 28 vs n = 43; Berer, n = 17 vs. n = 17; Cekanaviciute, n = 71 vs. n = 71. If a study reported a higher and lower relative abundances with the same taxonomy assignment, the cell was split into two colors, as was the case for Faecalibacterium prausnitzii. If a study reported a taxonomy assignment confidence score, only scores of 95% or better were included in the table. All taxa are statistically significant (P < 0.05) after FDR adjustment. NS, not significant; IMD, immunomodulatory drug; untr, IMD untreated; tr, IMD treated; MS, multiple sclerosis; NS, not significant.
247 OTUs (161 reduced in MS and 86 enriched in MS)
Different statistical tests were used to assess differences in OTU abundance. DESeq2 is a differential gene expression analysis based on the negative binomial distribution. Filtering is the process in which undesired (e.g., unreliable) OTUs are removed. Often when filtering is used, OTUs present in less than 5% of samples are discarded. In some cases, it was not possible to determine if the total OTU count was filtered or not before reporting. ANOVA, Analysis of variance; NA, not available.
). When IMD unexposed MS cases were compared to controls (Table 5), a higher relative abundance of the following taxa was found Methanobrevibacter (two studies, with a combined total of 37/60 cases/controls) (
We reviewed the recent literature (2008–2018) on the MS gut microbiota composition. Of the ten studies comparing the gut microbiota between MS participants and controls, the majority found no major differences in the overall composition of the gut microbiota in children or adults with MS relative to controls, as judged by alpha- or beta-diversity measures. Instead, subtle differences in the gut microbial communities were generally observed. At least two or more studies reported a higher relative abundance of Akkermansia and Methanobrevibacter and a lower relative abundance of Prevotella, Bacteroides (coprophilus and fragilis) and Faecalibacterium prausnitzii for MS cases relative to controls. Studies were generally too modest in size to adequately assess potential effect modifiers (confounders) such as drug exposure relevant to diversity or taxa-level findings.
4.1 Microbiota diversity
No study observed a significant difference in alpha-diversity between MS cases and controls and the majority found no differences in beta-diversity. However, the metrics and methods used varied across studies. While this makes comparisons challenging, the use of different diversity measures was not unexpected, as no single index perfectly summarizes local diversity (
). Beta-diversity specifically quantifies the variation in the taxonomic composition between samples; two of seven studies reported a significant difference between MS cases and controls (
Dysbiosis in the gut microbiota of patients with multiple sclerosis, with a striking depletion of species belonging to clostridia XIVa and IV clusters.
). However, it remains possible that findings were affected by IMD exposure; differences in beta-diversity were found in one study when IMD exposure was examined (
We identified several taxa that differed in their relative abundance between MS cases and controls across two or more studies, though their role in MS are largely unknown. Methanobrevibacter, an archaeal anaerobe and methanogen, was enriched in MS cases relative to controls (
), a condition which is common in MS. A. muciniphila was also enriched in MS cases relative to controls. A similar relationship was reported in Parkinson's disease (
). A. muciniphila has been shown to elicit a pro-inflammatory T lymphocyte response in vitro; however, in vivo studies using mouse models of MS have so far failed to elicit a similar reponse (
). Intriguingly, A. muciniphila may be beneficial in the setting of obesity or metabolic disorders, by supporting metabolic health and improving the intestinal barrier (
Akkermansia muciniphila inversely correlates with the onset of inflammation, altered adipose tissue metabolism and metabolic disorders during obesity in mice.
). These context-specific observations highlight the complexity of the gut microbiome, and a need to understand the underlying biology.
The remaining taxa identified—all of which are also common commensal bacteria of the human gastrointestinal tract—were all lower in relative abundance for MS cases relative to controls. Faecalibacterium prausnitzii is known for mitigating inflammation and may be depleted in the gut of individuals with other diseases, such as inflammatory bowel disease and irritable bowel syndrome (
Altered molecular signature of intestinal microbiota in irritable bowel syndrome patients compared with healthy controls: a systematic review and meta-analysis.
). These conflicting OTUs may actually be two different recently discovered phylogroups within the species F. prausnitzii which have recently been identified (
). The two phylogroups share 97% 16S rRNA gene sequence similarity but have different metabolic properties. Future studies planning to assign taxonomy may find it helpful to further classify F. prausnitzii into phylogroups when possible to better resolve the mapping contradiction and serve as a better discriminating biomarker (
). Interestingly, the relative abundance of Prevotella was lower in relapsing-remitting MS patients with ‘evidence of disease activity’ relative to those with ‘no evidence of disease activity’ in one study (
). Prevotella species were also present in the oral microbiota of new-onset rheumatoid arthritis participants, but not in controls, suggesting a possible role in this autoimmune inflammatory condition (
B. fragilis is thought to benefit human health by, for example, breaking down dietary fibers to produce short-chain fatty acids and anti-inflammatory polysaccharides (
). While there is limited relevant literature for Bacteroides coprophilus, this species merits further investigation for its potential role in MS.
Experimental studies allude to a pro-inflammatory MS gut microbiota. Neurological symptoms were exacerbated in animal models of MS when stool from individuals with MS was transplanted into the gut of mice with spontaneous or induced autoimmune encephalomyelitis, further supporting the association of the gut microbiota with MS (
Of the few confounding factors assessed to date in MS, exposure to disease-modifying drugs appears to be a likely candidate. However, most studies were too modest in size to formally assess the effect of confounders, hence much remains unknown. For example, at least 20 samples per group have been suggested in order to detect differences in taxonomic relative abundances (
). Exposure to several medications as well as stool consistency (a reflection of gut transit time) are considered important factors in explaining microbiome variation and are related to MS itself (
). A cross-sectional study assessing the differences between the gut microbiota of MS cases treated with either dimethyl fumarate (DMF; n = 33 cases) or glatiramer acetate (GA, n = 60) relative to IMD naïve cases (n = 75) was published in 2018, and although did not fulfill our inclusion criteria (due to absence of controls without MS), it is worthy of comment (
). Authors reported that MS cases exposed to either DMF or GA had lower relative abundances of the Lachnospiraceae and Veillonellaceae families compared to IMD naïve cases (
Constipation (reflecting a slow gut transit time) is common in MS, may influence the gut microbiota composition and contribute to a pro-inflammatory local environment. Microbiota from chronically constipated individuals was demonstrated to damage the intestinal barrier and further contribute to constipation (
). An enrichment of A. muciniphila and Methanobrevibacter spp. could be related to a slow gut transit and constipation, as shown in other conditions, including Parkinson disease (
). Methanogens may thrive in a gut with reduced motility, and may contribute to a slow colonic transit by augmenting methane production which acts as a neuromuscular transmitter, and has been shown to slow bowel movement (
). Understanding how these factors relate to MS and the gut microbiota's composition and function may clarify a possible causal role of the gut microbiota in MS. Alternatively, findings might point towards an opportunity to modify the gut microbiota to improve outcomes in MS. Future studies could assess stool consistency using proxy markers, such as the Bristol Stool Scale (
). The relationship between the IMDs used to treat MS and the gut microbiota is particularly intriguing and could provide additional mechanistic insights. Sufficiently powered, prospective longitudinal studies are needed to better understand the complex and likely dynamic relationship between MS, the gut microbiome, comorbidities, medication exposure, diet, and other lifestyle factors.
4.4 Heterogeneity of study design
Heterogeneity in the microbiota composition across studies may relate to differences in the sourcing of cases and controls and their characteristics, including: the broad age range of participants (from children to adults); (
)) and disease duration (which ranged from a few months to decades); host geographic location (although all studies were from largely westernized populations) and ethnicity (
). For example, although it is possible that individuals with a progressive disease course might differ in terms of their gut microbiota composition from those with a relapsing disease course, insufficient data and cases were available. Further, teasing apart the effects of age, disease duration, accrual of comorbidities and medication exposures from the underlying disease course would likely require access to a sizable cohort of individuals with progressive and relapsing MS.
Technical methods for quantifying and analyzing the gut microbiota also differed, including the choice of: quantification instrument, 16S rRNA sequences regions(s), and gastrointestinal tract site sampled. Computational methods also differed, including: the bioinformatics pipeline used to generate OTUs, OTU abundance normalization, and statistical tests employed. While the method of generating OTUs varied, all studies using an NGS platform generated OTUs at the same taxonomic resolution by clustering the 16S rRNA sequences based on a 97% similarity threshold.
4.5 Strengths and limitations
The strengths of this review include its systematic, reproducible approach, and pre-registered protocol. Our systematic review also provides insights into the heterogeneity in microbiome study design including an overview of the differences in the computational pipelines. However, all studies included were relatively modest in size, with the total number of available data pertaining to 286 MS cases and 296 controls from a limited number of regions in the world. It remains possible that associations have been missed, particularly with the lower abundant taxa. For simplicity, we only reported findings on taxa that were similarly observed across two or more studies. Further, all studies were considered together, including one which sampled duodenal mucosal tissue. It remains possible that different physiological niches in the gastrointestinal tract will harbor distinct microbiota communities of relevance in MS. Interrogation of the gut microbiota in MS was primarily conducted using 16S rRNA sequencing which is typically unable to assign taxonomy below species level and is incomplete at low taxonomic ranks. It was not possible to match and compare individual OTUs identified across studies. We found no published study investigating the virome or mycobiome (fungi microbiome) in MS.
4.6 Concluding remarks
To our knowledge, this article is the first to systematically review the scientific literature investigation the link between the gut microbiome and MS. Despite the modest cohort sizes, diversity in the geographical location of participants and sample processing and bioinformatics pipelines used, consistent patterns are emerging: several taxa were similarly identified as being over or underrepresented in MS versus controls. A better understanding of a possible causal role of the microbiota in either facilitating the onset of MS, or outcomes in MS, including perpetuating comorbidities will facilitate our ability to harness the microbiome to affect positive change in MS.
5. Addendum
5.1 Summary of articles published between June 20th 2018 and August 24th, 2019
The systematic literature search was updated on August 24th, 2019 by one reviewer (AM). Of 122 new articles identified, three fulfilled the inclusion criteria, one from the US, one from Canada, and one from China (
Briefly, the US-based study included stool samples from 25 RRMS cases, all unexposed to IMDs or corticosteroids 3 months prior to stool collection (80% were women, mean age = 44.0 years), and 24 controls (12.5% women, mean age = 49.3 years). All bacteria as well as the spore-forming bacterial fractions alone were sequenced at the 16S rRNA V4 region, using the Illumina NextSeq platform and clustered into closed-reference OTUs (97% sequence similarity). No differences were observed in alpha-diversity (Chao1) or beta-diversity (unweighted UniFrac) between cases and controls for either the spore-forming bacteria or total bacteria. Differences between MS cases and controls in the relative abundance of OTUs of spore-forming bacteria fractions including: lower Ruminococcus gnavus, Ruminococcus bromii, Veillonella dispar, and a higher relative abundance of Propionibacterium acnes, Staphylococcus epidermidis, Clostridium perfringens, and Clostridium citroniae were reported (all P < 0.05). Taxa-level differences between cases and controls among total bacteria and the effects of possible confounders were not reported (
The Canada-based study included stool samples from 19 MS cases (average age = 47.3 years; 14 were women; MS course was not reported) and 23 healthy controls (average age = 32.4 years; 12 were women). In addition, the authors included 20 Crohn's disease cases, 19 ulcerative colitis cases, and 21 rheumatoid arthritis cases (not reported here). DNA extracts were sequenced at the 16S rRNA V4 region using the Illumina MiSeq platform, clustered into de novo OTUs (97% sequence similarity). Because the dataset had an unusually low abundance of Gram-negative bacteria, the differential taxa abundance testing included only OTUs from Gram-positives phyla (i.e., OTUs within the phyla Firmicutes, Actinobacteria, and Tenericutes). The author's main goal was to compare across all the immune-mediated diseases relative to controls. Here, we report only the comparisons between the MS cases and controls. Alpha-diversity (i.e., Chao1, ACE, Shannon index, and Simpson diversity index, based on all phyla) did not differ between MS cases and controls. Beta-diversity was not directly compared between MS cases and controls. Significant differences between MS cases and healthy controls in the relative abundance of gram-positive bacteria at the genera level included: Lower Butyricicoccus, Dialister, Faecalibacterium (consistent with a previous finding (
Dysbiosis in the gut microbiota of patients with multiple sclerosis, with a striking depletion of species belonging to clostridia XIVa and IV clusters.
)), Fusicatenibacter, Gemmiger, Lachnospira, Sporobacter, and Subdoligranulum in MS cases relative to controls and higher Actinomyces, Eggerthella, Anaerofustis, Clostridium group III, Clostridium group XlVa, Clostridium sensu stricto, Faecalicoccus, Streptococcus and Turicibacter in MS cases relative to healthy controls (P < 0.05, Kruskal–Wallis test and Dunn's post hoc tests for multiple comparisons, with FDR correction) (
The third study included stool samples from 34 RRMS cases (21 unexposed and 13 exposed to an immunosuppressant [e.g., azathioprine, methotrexate, or mycophenolate]) and 34 healthy controls from China (in addition to 34 individuals with neuromyelitis optica spectrum disorder). DNA extracts were sequenced at the 16S rRNA V3-V4 region using the Illumina MiSeq platform and clustered into de novo OTUs (97% sequence similarity). Alpha-diversity (i.e., Chao1, Shannon index, and Simpson diversity index) did not differ between all MS cases and healthy controls. Although beta-diversity did differ for both exposed MS cases and unexposed MS cases compared to healthy controls, (using a non-phylogenetic gain distance metric, P < 0.01 and P < 0.05, respectively; Mann−Whitney U test), the authors used an unconventional approach, by comparing distance values computed from the first principal coordinate of a principal coordinate analysis. The relative abundance of Prevotella was lower for the MS cases compared to controls, consistent with findings from prior studies in MS (exposed MS cases vs. controls, P < 0.05 and unexposed MS cases vs. controls, P < 0.0001; Mann−Whitney U test). As in the prior Canadian study (
), the relative abundance of Streptococcus was higher for the MS cases compared to controls, (exposed MS cases vs. controls, P < 0.001 and unexposed MS cases vs. controls, P < 0.0001) (
The new articles support the conclusion that there is no major differences in the overall composition of the gut microbiota in individuals with MS relative to controls, as assessed using alpha- or beta-diversity measures. Two studies consistently identified two taxa that differed in their relative abundance between MS cases and controls: lower Faecalibacterium and higher Streptococcus in MS cases relative to controls (
). The former observation concurs with a finding from a study in our systematic review, the latter represents a ‘new’ observation.
Author contributions
AM and HT contributed to conception and design of the study, acquisition, analysis and interpretation of data, and drafted the first version of the manuscript, tables and figures. JF, FZ, CB, GV, MG, and EW contributed to acquisition and interpretation of data and provided critical review of the manuscript.
Potential conflict of interest
HT is the Canada Research Chair for Neuroepidemiology and Multiple Sclerosis. Current research support received from the National Multiple Sclerosis Society, the Canadian Institutes of Health Research, the Multiple Sclerosis Society of Canada and the Multiple Sclerosis Scientific Research Foundation. In addition, in the last five years, has received research support from the Multiple Sclerosis Society of Canada (Don Paty Career Development Award); the Michael Smith Foundation for Health Research (Scholar Award) and the UK MS Trust; speaker honoraria and/or travel expenses to attend CME conferences from the Consortium of MS Centres (2013, 2018), the National MS Society (2014, 2016, 2018), ECTRIMS (2013, 2014, 2015, 2016, 2017, 2018, 2019), Biogen Idec (2014), American Academy of Neurology (2013, 2014, 2015, 2016, 2019). All speaker honoraria are either declined or donated to an MS charity or to an unrestricted grant for use by HT's research group.
AM, FZ and JF were funded through research grants held by HT. AM received funding from the Multiple Sclerosis Society of Canada.
CB, GV, MG, and EW have no conflict of interest to report.
Acknowledgements
The study authors received funding from The Multiple Sclerosis Scientific and Research Foundation, EGID: 2636 (PI: Tremlett). Ali Mirza was supported by an endMS Doctoral Studentship Award from the Multiple Sclerosis Society of Canada, EGID: 3246.
Appendix 1. Information retrieved and extracted from each included study
Tabled
1
1. Publication information: name of author(s) and year of publication,
2. Study information: study design, geographic location of recruits (e.g., National Center of Neurology and Psychiatry Hospital, Tokyo, Japan), MS diagnosis criteria (e.g., McDonald criteria), and eligibility criteria, and enrolment period.
3. Case-control demographics and characteristics close to the time of microbiota sample collection: age, sex, race/ethnicity, MS disease course (e.g., relapsing-remitting onset MS or primary progressive), case or control status, MS disease duration, disability level (e.g., the Expanded Disability Status Scale (EDSS) score), lifestyle factors, as available (e.g., BMI, diet-related metrics). Medication use, including drugs used to treat MS, e.g., the immunomodulatory drug (IMDs), captured as therapeutic class or generic name (e.g., interferon-beta, glatiramer acetate) and IMD treatment status [treated/untreated/naïve (as defined by the study authors)]. Systemic corticosteroid around the time of sample collection (yes, no). Other medication use, including antibiotics.
4. Microbiota sampling and quantification information: body site sampled (e.g., stool, mucosa biopsy of small intestines, etc.), number of samples collected, number of samples excluded, DNA isolation kit, microbiota quantification instrument, sequence molecule/ region, primer sequence, method of generating OTUs (‘Operational Taxonomic Units’; species-like taxonomy), and OTU statistical normalization method.
5. Microbiota analysis results: Total OTUs, differential abundance statistical method, taxa-level [phylum, genus, species (or individual OTUs)] differences between cases and control, number OTUs differentially abundant before and after multiple-testing adjustment, potential confounding factors considered in the study design and/or adjusted statistically in the analyses, diversity metric(s) considered, diversity comparisons between groups, e.g., MS vs. controls, and when available, untreated MS vs controls and treated MS vs untreated MS.
Dysbiosis in the gut microbiota of patients with multiple sclerosis, with a striking depletion of species belonging to clostridia XIVa and IV clusters.
Akkermansia muciniphila inversely correlates with the onset of inflammation, altered adipose tissue metabolism and metabolic disorders during obesity in mice.
Altered molecular signature of intestinal microbiota in irritable bowel syndrome patients compared with healthy controls: a systematic review and meta-analysis.
30407-9&id=gr1.jpg){kind=link}