A systematic review of enteric dysbiosis in chronic fatigue syndrome/myalgic encephalomyelitis

Four databases were searched: EMBASE, Medline (EBSCOhost), Pubmed and Scopus. The following terms were systematically searched as full-text and Medical Subject Headings (MeSH) terms: chronic fatigue (which includes chronic fatigue syndrome and myalgic encephalomyelitis) or systemic exertion intolerance and microbiome (which includes microbiota, gut bacteria/flora, intestinal bacteria/flora, enteric bacteria/flora, microbial flora and microflora), commensal or dysbiosis. All search results were limited to publication date since the establishment of the Fukuda criteria (year 1994–2018). The primary search was performed on 22nd February 2018, and the final search was completed on 31st March 2018.

Studies that fulfilled the following criteria represented in their titles or abstracts were eligible for inclusion: (i) studies that were conducted in humans; (ii) studies written in English available as full text through institutional access; (iii) all studies that investigated the bacterial composition of the microbiome in CFS/ME subjects; (iv) CFS/ME diagnosis according to Fukuda (1994), CCC (2003) or ICC (2011); (v) adults aged 18 years and over; (vi) year searched 1994 to the present year to exclude earlier articles prior to establishment of Fukuda criteria; and (vii) journal articles reporting studies based on original research.

Studies were not included in this review if less than two key search terms were not stated in the title or abstract, and if the criteria used to diagnose CFS/ME were unclear following screening of the full-text. Additionally, studies that used other patient groups (e.g. fibromyalgia (FM), IBD) as a comparison to the CFS/ME cohort were excluded as these conditions commonly co-occur with CFS/ME. Duplicate studies, case reports/studies, or review articles and studies not meeting the above inclusion criteria were also excluded. The primary outcome of interest for this review was the composition of the gut microbiome in CFS/ME. Secondary outcomes evaluated were the association of gut microbiota composition with indicators of illness severity, such as QoL, physical activity and psychological wellbeing. Studies were also excluded if they combined CFS/ME with other patient groups [e.g. CFS/ME and irritable bowel syndrome (IBS)]. Although CFS/ME often co-occurs with IBS, the co-occurrence of IBS was excluded to reduce the risk of confounding, as gut microbiota composition is altered in IBS.

Titles and abstracts for each article were initially screened on the basis of eligibility criteria. Full-text articles and study quality were independently assessed by two review authors for suitability for inclusion in this review. This was later reassessed and confirmed by all other team members. Eligible studies were read, and the relevant data were extracted (Tables 1 and 2) including (i) study design, (ii) CFS/ME case definition, (iii) country, (iv) sample size, (v) age of participants, (vi) sex and percentage of female participants within the group, (vii) illness duration, (viii) body mass index and weight, (ix) method of quantifying the gut microbiome composition and (x) result of investigation and level of statistical significance.

The Downs and Black checklist was used to assess study quality and bias [20]. Score ranges have previously been categorised as excellent, good, fair or poor to assist interpretation of scores [21]. For the purposes of this systematic review, these have been converted to percentage ranges, as items for assessing study quality were not consistently relevant across all studies included. The scores are as follows: excellent (94–100%), good (72–94%), fair (54–72%) and poor (< 54%).

Figure 1 presents the PRISMA flow diagram with the number of included and excluded studies. A total of seven studies were included in this systematic review of the microbiome composition in CFS/ME patients and are summarised in Table 2. The included studies were one randomised control trial (RCT) and six observational cohort studies/case control studies. Studies varied in study quality, with the Downs and Black checklist scores ranging from 47 to 80% (Table 1). Of the included studies, three out of seven were scored as good quality (72–94%) [22‐24] and two were scored as fair quality (54–72%) [25, 26]. The remaining two were scored as poor quality (< 54%) [27, 28].

The participant characteristics of the included studies are summarised in Tables 2 and 3. Diagnosis of CFS/ME was made using the Fukuda (1994) case definition in four studies [22, 25, 27, 28] and the CCC (2003) in two studies [24, 27]. The remaining study used both Fukuda and CCC as well as Holmes (1988) [23]. Due to the similarities between the Holmes [29] and Fukuda criteria, in addition to the CCC criterion, this publication was included in this review. The mean sample size for each study was 89 participants. The primary outcome of this review was gut microbiome composition in CFS/ME, which was reported in all seven studies [22‐28]. All used patient faecal samples as a proxy for investigating intestinal flora composition. Of these, three used culturing methods [23, 24, 26], with one using matrix-assisted laser desorption ionisation time-of flight mass spectrometry (MALDI-TOF MS) after culturing to identify genera [24], three used 16S ribonucleic acid (RNA) amplification and sequencing [22, 25, 27] to investigate bacterial flora of the gut and one study used 18S deoxyribonucleic acid (DNA) amplification and sequencing [28] for investigating eukaryotes in the microbiome. Secondary outcomes evaluated were GI symptoms, psychological wellbeing, cognitive function, QoL and pain and fatigue scores. Four of the included studies did not investigate secondary outcomes as part of their methodology [23‐25, 27]. One study reported GI symptoms using an unknown tool and the Bell’s Disability Scale to measure severity of fatigue and cognitive symptoms [28]; two reported psychological symptoms using the Multidimensional Fatigue Inventory (MFI) and profile of mood states (POMS) [22], or Beck Anxiety Inventory (BAI) and Beck Depression Inventory (BDI) [26]; and one reported cognitive function, QoL measures and pain and fatigue scores using MFI and POMS [22].

Of the included studies, six reported differences in the microbiome in CFS/ME compared with healthy control (HC) [22‐25, 27, 28]. One study reported changes to the microbiome following administration of a probiotic intervention compared with a placebo control as part of a double-blind RCT [26] (Table 4). The following significant differences were reported with respect to CFS/ME patients relative to control subjects: increased Clostridium spp. (p = 0.020), and decreased total bacteria (p = 0.005), total anaerobic bacteria (p = 0.021) and Bacteroides spp. (p = 0.009, Bacteroides vulgatus and Bacteroides uniformis (not significant) [24]; increased Lactinofactor (p < 0.001) and Alistipes (p < 0.05), and decreased Roseburia (p < 0.05), Syntrophococcus (p < 0.05), Holdmenia (p < 0.01) and Dialister (p < 0.05) in the Norwegian population, and increased Lactinofactor (p < 0.01) and decreased Asaccharobacter (p < 0.05) in the Belgian population [25]; reduced phylogenetic diversity (p = 0.004) [27]; increased total aerobes (p < 0.001), Enterococcus faecalis (p < 0.001), Streptococcus sanguinis (p < 0.001) and gram-negative species (p < 0.01) [23]; and decreased mean relative abundance of Actinobacteria (p < 0.05) [22]. Although other differences were reported in these studies, not one was statistically significant.

From the seven studies, three investigated secondary outcomes on one or more of the following parameters: GI symptoms, psychological wellbeing, cognitive functioning, QoL and pain and fatigue scores (Table 5). One study reported that increased levels of Bifidobacteria and Lactobacillus in faecal samples of CFS/ME patients receiving 8 weeks of probiotic supplementation were associated with a significant improvement in anxiety scores (p = 0.01), but not depression scores. [26]. The other study observed a decrease in the mean relative abundance in Actinobacteria in CFS/ME patients, which corresponded with higher MFI scores in CFS/ME vs. HC (p < 0.05), particularly greater general and physical fatigue, reduced activity and motivation and greater mental fatigue [22]. These findings may act as a possible link between microbiome composition and CFS/ME symptom severity and other outcomes.

The present review was comprised primarily of female participants ageing between 35 and 54 years. These findings are consistent with epidemiological studies of CFS/ME that report a greater prevalence of the illness in females and those aged 35 and 45 years [16, 36, 37]. The participants resided in Australia, Europe or North America, and one of the seven publications included information regarding the ethnicities of participants [25]. Moreover, most of the studies examined in the current review used the Fukuda (1994) case definition [5]. The broad nature of the Fukuda case definition may have contributed to the inconsistency of the findings generated by studies in the present review to identify notable alterations to the microbiome of CFS/ME patients. Revised criteria after Fukuda (1994), such as CCC (2003) and ICC (2011), better defined the illness that characteristically presents with a diverse array of symptoms and creates consistency with CFS/ME diagnosis. Similarly, not all studies used the same case definition of CFS/ME, making it difficult to compare findings due to the inherent differences between the criteria.

A number of methods were used to identify microbiota alterations in the gut, with no two studies in this review using identical methods of investigating the microbiome. This finding suggests that there is currently no standardised protocol for investigating the composition of the microbiome, which limits the ability to make appropriate comparisons between studies. Furthermore, the instruments to measure secondary outcomes (e.g. fatigue, psychological symptoms) were inconsistent between the studies that considered these parameters.

Although the studies recruited patients through clinics, universities or hospitals, none reported the severity spectrum of illness presentation. Due to the nature of recruiting, many of these studies may have excluded severely affected CFS/ME patients that were rendered house- or bed-bound due to their inability to attend an out of home location for screening or sample collection. This potential sampling bias is likely to confound the results and limit the validity of the findings to only patients with mild or moderate illness severity with the capacity to venture outside their homes. It is therefore essential for future research to consider the varying severity present in CFS/ME when designing and implementing research methodology to accommodate and include all representations of this illness under a range of settings (e.g. in a clinic or patient’s home).

CFS/ME is complex illness believed to be a multisystemic disorder affecting the immune, nervous, GI, cardiovascular and endocrine systems [38]. The pathomechanism of this condition is yet to be established; however, it has been suggested that CFS/ME is a manifestation of gut dysbiosis [30, 31]. A systematic review published in 2016 on the use of certain drug therapies, which included antibiotics, did not show these therapies to be beneficial in treating CFS/ME [39]. Similarly, a systematic review published earlier this year concluded that probiotics were ineffective in treating CFS/ME [4]. The findings of these systematic reviews suggest that alterations in the gut microbiome do not contribute to the pathomechanism of the illness. Consequently, administration of probiotics or antibiotics as a means to treat CFS/ME is not supported by the evidence.

Each observational study, while using different research methods, reported variability in the gut microbiome composition of CFS/ME patients compared to HC; however, only some studies reached statistical significance. Importantly, all studies included in this review used faecal samples as a proxy to determine gut microbiome composition. The use of stool bacteria may merely represent luminal bacteria and not gut mucosal flora analysed by gastric tissue biopsy [40]. Additionally, the use of culture methods which are less specific than DNA and/or RNA amplification and sequencing techniques may be problematic [41, 42]. Increased total anaerobic bacteria or decreased total aerobic bacteria was described in two studies, both using culturing methods and nuclear magnetic resonance (NMR) spectroscopy [23, 24]. Although increases in Clostridium spp. and Enterococcus faecalis were reported [23, 24], diets that are high in sugar and high in fat are believed to encourage growth of Clostridium and Enterococcus spp., among others [43]. These two studies did not control for diet ahead of sample collection, thereby limiting the validity of their findings. This suggests the necessity to control extrinsic factors, such as diet, in addition to utilising more sensitive gut microbiota profiling methods.

The studies that surveyed secondary outcome measures related to the symptoms of CFS/ME when compared with HC reported that CFS/ME patients are more commonly affected by GI disturbances; higher disability, pain and fatigue scores; and reduced emotional wellbeing, motivation and mental functioning [44]. Again, inconsistency in the instruments used to measure these parameters made it difficult to perform valid assessments of the data. A paper not included in this review due to the inclusion of patients with IBS by Nagy-Szakal and colleagues reported on direct correlations between the relative abundance of specific bacterial strains and scores on the MFI and 36-Item Short Form Health Survey (SF-36) questionnaires [45]. The RCT conducted by Rao et al. that administered 24 billion colony forming units Lactobacillus casei strain Shirota to the treatment group of CFS/ME reported significant increases in Bifidobacteria and Lactobacillus compared to the control group after 8 weeks [26].

The quality of the studies identified and included in this review ranged from poor to good. The Downs and Black Checklist was used to assess the quality of studies included in this review because it has previously been identified as a reliable tool for assessing both case-control and RCT studies, which have been examined in this systematic review [46]. Furthermore, questions not relevant to the current review from this comprehensive checklist could be eliminated without significantly impacting its ability to differentiate the overall quality of the studies and their findings. Using this tool, we were able to illustrate that the standard of current evidence for the existence of enteric dysbiosis in CFS/ME is inadequate to justify its inclusion as a criterion for diagnosis or basis for treatment.