The Gut Microbiome in Myalgic Encephalomyelitis (ME)/Chronic Fatigue Syndrome (CFS)

ME/CFS has a convoluted definition history, creating confusion by changing terminology and classification in the past (41–43). The disease Chronic Fatigue Syndrome (CFS) was first described in 1988 to propose a new name for the chronic Epstein-Barr virus syndrome (44). In 1994, CFS was redefined by the Fukuda Definition with the criteria of at least six months of unexplained chronic fatigue and requiring at least four symptoms out of eight listed (45). The Fukuda Definition criteria are still frequently used in research.

Years before CFS diagnostic criteria were introduced, the first formal description of Myalgic Encephalomyelopathy (ME) appeared in a Lancet editorial in 1956 after an illness outbreak (46, 47). Later the cardinal symptoms of the neuromuscular disease were described as prolonged muscle weakness, neurological dysfunction and the involvement of other body systems. Muscular fatigue occurs after minimal physical exertion and sensory, cognitive and autonomic functions are impaired (41).

In 2003, a more differentiated disease definition was introduced for both diseases with the Canadian Consensus Criteria. They described the mixed condition ME/CFS with a detailed diagnostic scheme with specifications of immunological, neurocognitive and other impairments and a demarcation to psychiatric diseases (48, 49). Subsequently, researchers increasingly considered CFS and ME to be similar diseases and started applying the hybrid diagnosis ME/CFS. Due to inconsistent use of clinical criteria in research studies, an expert group was formed to develop improved diagnostic criteria for CFS and ME, resulting in an International Consensus Criteria definition (41, 50). According to the consensus report that was based on a study with over 2500 patients, waiting six months for a diagnosis was no longer recommended as a requirement. Fatigue as the cardinal symptom was replaced by post-exertional malaise (PEM). Eight symptoms from four areas and a minimum of a 50% reduction from pre-illness activity capacity are required for diagnosis under this definition (50). In 2015, the US National Academy of Medicine (at that time “Institute of Medicine”) recommended, with little success, to change the hybrid ME/CFS into the new name Systemic Exertion Intolerance Disease (1).

However, several researchers note that ME and CFS originally did not share the same disease definition and perhaps should not be conflated (41, 49, 51). On one side, ME describes more clearly impaired individuals and there is not yet highly objective evidence for inflammation in the brain, which makes the term ME confusing (1, 2, 51, 52), and on the other side, the term CFS often results in stigmatization and trivialization, which does not do justice to the serious complaints and impairments (1, 52). Nevertheless, in practice and research, the hybrid ME/CFS has become a predominantly used term due to the overlapping clinical picture (1, 41, 51). We therefore also use ME/CFS in the present work.

ME/CFS is a severe multisystem disease with a high degree of physical disability that can lead to a completely bedridden condition with a high need for care. The disease is characterized by debilitating fatigue with unrefreshing sleep, neurocognitive impairments and flu-like symptoms such as muscle weakness and pain, headaches, sore throat and tender lymph nodes, among others. The malaise and the accompanying symptoms are worsening dramatically after minimal physical, orthostatic and cognitive activity, which is referred to PEM. PEM is the cardinal symptom and mostly comes with a delay of 24 hours after the overexertion leading to a significant reduction (at least 50%) of the patient’s energy and activity level. The recovery period of such an exhaustion usually takes more than 24 hours and can be prolonged to weeks. The course of the disease is fluctuating, the severity of symptoms can change drastically only in a few days. Patients often describe allergies, intolerance in any sort of stimulation such as temperature extremes, light, noise, specific odors or chemicals and simple conversations (2, 45, 50, 53). Patients often suffer from gastrointestinal symptoms, including constipation, diarrhea and intestinal discomfort (53). There is a frequent comorbidity of ME/CFS with irritable bowel syndrome (IBS) ranging from 38% (54) to 42% (12) and up to 92% (8). A trial with 48 patients indicates that over 70% of patients have gastrointestinal symptoms (13). The medication intake against digestive or gastrointestinal problems is over 35% (54). At least 25% of the patients are house- or bedbound (2, 55, 56), representing the moderate and severe forms of ME/CFS (50). According to a large epidemiological study, almost half of the patients are incapable of working (54). A previous study reported 87% as unable to work (56). 48% of patients are not able to pursue any productive activity during their worst periods (54). Severely affected patients describe their limits with restricting their communications and lying in darkness in bed, not being able to shower regularly, or even listen to music or watch TV (57). The illness, which can be as disabling as multiple sclerosis (MS) or systemic lupus erythematosus (SLE) (58), is defined by the WHO in the ICD-10 as a disorder of the nervous system (59). However, taking into account the fact that there still exist different assumptions about the pathogenesis and etiology, it is debated whether this is the right classification (2, 41, 49).

The global prevalence of ME/CFS depends on which disease definition is used and ranges widely in the literature, also because of the lack of objective diagnostic tests (2, 53, 60). The National Institute of Health and Care Excellence (NICE) reported an estimated number of 0.2% to 0.4% of the world’s population suffering from ME/CFS (58). A meta-analysis calculated the prevalence globally as high as 0.8% based on clinical assessments and 3.3% based on self-assessments (61). In 2019 Valdez and co-workers confirmed that ME/CFS is not a rare disease estimating the prevalence rate for ME/CFS to be two times higher than the current estimates at 857 per 100’000 individuals (0.86%) using machine learning techniques (62). ME/CFS affects more females than males (2) by a factor of approximately 1.5 to 2 (63).

The economic burden on society is high, with an estimated $7 billion in direct annual costs only in the US (64). Compared to MS or SLE, the costs are around 50% higher for ME/CFS (62). Patients diagnosed with ME/CFS need more health care resources than the rest of the population before their first recorded diagnosis. Patients often consult several doctors for as long as ten years until a diagnosis is made (65). According to a Danish study in 2015, ME/CFS patients have a lower health-related quality of life score (-0.29 in the EQ-5D-3L score) than patients with MS, cancer or stroke (66). A recently published cross-sectional study confirms that patients with ME/CFS are more disabled than patients with MS in almost all areas, including also with regards to employment and income (67).

ME/CFS usually tends to develop in the middle years of life (68), typically with two different onset patterns: a sudden onset that patients can usually remember or a more gradual onset with a slow worsening. A majority of the patients reported having been healthy before, fully functional and with an active social life (54, 69, 70). A cohort study in the United States in 2019 reported that most patients experienced infectious episodes before the disease onset (64%), followed by stressful incidents (39%) and exposure to environmental toxins (20%), from which they never recovered (54). These findings are consistent with earlier studies that also reported the exceptionally high frequency of an infection-associated onset or of co-infections in ME/CFS patients (69, 71, 72). Although several research teams were able to show that a chronic bacterial or viral infection exists in a subset of ME/CFS patients (72–75), there is no clear evidence that infection is the cause of disease or of any pathogen uniquely involved (17). However, some reports of clustered outbreaks of ME/CFS after epidemics (76) raise the question about specific pathogens more frequently triggering ME/CFS (46, 77).

The severity of fatigue and the symptom constellation can change over time (78). The long-term development of the disease varies from patient to patient (69), but crucially a large number of patients (70 – 90%) reported no improvement over time (54). Although a few patients can show a temporary remission (69), a spontaneous full recovery from the disease is extremely rare occurring in only 5% of patients (79).

Currently, there is no diagnostic test or validated biomarker for ME/CFS available (2, 80), although some results give hope for a future diagnostic biomarker or the use of an immunosignature assay (13, 81–85). The diagnosis is based on the Canadian Consensus Criteria recommended for research studies (1, 50, 53, 78). As mentioned, PEM is the most distinct symptom for differentiating ME/CFS from other diseases with chronic fatigue (53, 70). A meta-analysis reported PEM 10.4 times more common with a ME/CFS diagnosis than in healthy controls (86). Patients often describe these cognitive and physical exacerbations as “crashes” or “collapses” (57). The worsening after physical activity or even normal daily activities also helps to exclude depression or other diseases, because in most other diseases physical activity normally leads to an improvement of fatigue (87, 88). Disease onset with an infection is another important clue to diagnosis (78).

Even though ME/CFS is a common disease, it is often ignored in medical education and research. Various studies show that around 80% of the patients are not diagnosed correctly due to deficits in medical education (53, 89). This is a major challenge not only for the patients, who have to deal with inexperienced care providers, but also for society and the public health system (2, 64).

Some recently published studies suggested that a diagnostic tool may become available in future, as in several cases ME/CFS patients could be reliably distinguished from controls using platforms including nano-electronic blood-based assays (83), CD8+ T cell analysis (81), fecal metagenomic profiles (12) and metabolite analysis (12, 81, 83, 84).

The underlying etiology or a specific cause of ME/CFS is currently not known. However, over the past years multiple studies have shown elementary evidence of a multi-factorial etiology involving neuro-immuno-endocrinological patterns, metabolic alterations and also genetic components (53, 90–92). Here, we focus on the findings of a dysfunctional immune system, impairments in energy metabolism, genetic predispositions and neurological abnormalities. Finally, a possible role of the gut and its microbiome will be discussed as a current hypothesis.

There is currently no pharmacological or non-pharmacological therapy available with clear evidence for the management of ME/CFS (2, 70, 138, 139). Pacing, the careful dosing of all types of activities as an energy self-management method, is one of the most recommended strategies to avoid relapses due to overexertion. But it is not curative. Patients should learn to read their fluctuating symptoms, pay attention to early signs of PEM and trust their own experience. They have to learn to manage their cognitive, physical, emotional and social activities to reduce a worsening of their symptoms (2, 53, 54, 70, 86). But this is difficult, because the overexertion normally is not recognized immediately and the avoidance of increasing the activity level while feeling well is a major restriction, limiting compliance (57). General treatment should be symptom-based (78), for example wearing eyeshades against light sensitivity or headphones that protect the patient from sound and noise sensitivity (6). Over half of patients use a variety of medications to manage sleep, pain and endocrinological problems and to cope with mood and gastrointestinal issues (54). Among them are also many drugs that are off-label, experimental or non-prescription (57). The new published clinical guideline of NICE warns that therapy should not represent a fixed increase of any kind of activity, because this may lead to a worsening of symptoms (2).

Few treatment modalities targeting autoimmunity have been evaluated in clinical trials in ME/CFS so far showing some evidence for efficacy of rituximab, cyclophosphamide, and IgG in a subset of patients (93, 107). Removing IgG from plasma by immunoadsorption recently showed first evidence of efficacy (110).

In some experts’ perspective, the microbiome with all its genetic information, metabolism, physiology and pathophysiology as well as transplantability should be considered a separate organ (140–142). Considering that the human body contains the same numbers of human and bacterial cells, this is not surprising. Of the overall body weight, about 0.3% is composed of bacteria and the highest density of microorganisms can be found in the digestive tract (143). The term microbiota encompasses the totality of all bacteria and other microorganisms like fungi, protozoa and viruses contained in the body and on the skin (142, 144).. The microbiome is defined as the community of these microorganisms (microbiota) in their well-defined habitat and also refers to their structural elements, metabolites, genetic information, and their surrounding environmental conditions. It is also described as a dynamic interactive micro-ecosystem, which is integrated in the macro-ecosystem where it has crucial functions for health (144). We focus here on the composition and the function of the microbiome in the gut, which is extremely diverse and different in every human, influenced by eating habits, other environmental exposures, sleep, stress, genetic factors and medical treatments (4, 5). Microbial colonization begins immediately after birth, so that the mode of delivery (vaginal or via elective Caesarean (C-) section) has an effect on the establishment of the microbiome (145). C-section is accompanied by reduced bacterial diversity in the gut, an altered development of the immune system, and increased allergies and asthma (146). After birth, how an infant is fed (with oral to skin contact during breastfeeding or with more distance as with formula) has profound effects on the early microbial colonization. In addition, breastfeeding affects the neonatal microbiome both directly through the milk microbiota as well as indirectly through shaping of the neonatal immune system through ingredients such as human milk oligosaccharides, secretory IgA, antimicrobial proteins (AMPs) and tryptophan metabolites (147). The composition of microbiota is already relatively adult-like and stable after the first three years of life. Maternal factors, birth mode, the exposure of postnatal factors like antibiotics, breastfeeding and the environment shape or disrupt the ecosystem of the neonatal microbiome in this early phase and can influence the development of the microbiome and the immune system (145).

The gut microbiome as a complex has multiple functions. One important function is the energy extraction from food by fermenting carbohydrates resulting in the production of various compounds for the host (148) like the short-chain fatty acids (SCFAs) butyrate, acetate and propionate (149) and amino acids like γ-aminobutyric acid (GABA) and vitamins (150). Intestinal microorganisms not only play a role in the digestive process and as an important barrier against pathogens (i.e. colonization resistance), but also train the immune system resulting in the production of antibodies and secretion of AMPs. AMPs, toll-like-receptors, innate lymphocytes and IgA-production are all developing in the epithelial cells after gut colonization in a newborn (11). The gut epithelium recognizes signals from the immune cells in the lamina propria and serves as a mediator between intestinal microbiota and the immune system. The recognition of pathogen-associated molecular patterns (PAMPs) of microorganisms, for example lipopolysaccharide (LPS), is ensured by receptors such as TLR, NOD, and PRR (5). Other important functions are metabolism, the production of vitamins, steroid hormones and neurotransmitters (151) and the modulation of mood, cognition and memory through various mechanisms forming a “gut-brain axis” (152, 153). Overstimulation through bacterial dysbiosis, LPS or an increased gut permeability can lead to systemic inflammation (153).

There are beneficial disease-preventing bacteria like Bifidobacteria or Lactobacilli, but also bacteria that are associated with disease development (13, 154). For a long time neglected, studies of the microbiome have recently become more popular with the advent of new molecular methods, efficient computational pipelines and enhanced reference databases to investigate its composition (7, 155). These new methods allow to search for specific disease biomarkers and evaluate treatment options by influencing the microbiome. A large number of studies reported that shifts in the microbiota towards a loss of health-associated bacteria and an accumulation of pathogenic bacteria are associated with metabolic and immune impairments and several disease states including susceptibility to infectious diseases, obesity and metabolic problems, liver diseases, neurological diseases and cancer (5, 156–158). The phyla Firmicutes and Bacteroidetes make up 90% of the gut community, and their ratio has been linked to intestinal and host metabolic health, though not always consistent between studies. Gut ecosystem gene richness (i.e. the number of unique genes) and taxonomic diversity (i.e. the number of bacterial taxa represented) have been strongly linked to gut health and may indicate better resistance to external stress factors. Both measures decrease with age (145, 150, 158). A higher intestinal microbial diversity is also associated with a better host fitness (159), thicker intestinal mucosa and increased production of AMPs. Also, capacity to produce different SCFAs including butyrate and propionate is associated with healthy intestinal composition. For example, in metabolic disorders the production of butyrate and propionate is low (5).

A “perfect” microbiotal state is not recognized, as intestinal homeostasis varies strongly even in health (158, 160, 161). It is also challenging to attribute bacterial markers to single diseases because these trends often overlap and because of the dynamic nature of these findings. Thus some researchers discourage the use of the arguably inconsistent term “dysbiosis” and warn against inappropriate conclusions from research results (5, 157). Despite the need for such caution, repeated findings of microbiome alteration common to diseases sharing pathomechanisms indicate further insights that could be systematically investigated.

Intestinal microbiota alterations and dysbiosis have already been detected in several ME/CFS studies (12–16), but a specific consistent microbial signature has not been found yet (13, 17). The overview of microbiome studies in ME/CFS patients shows an inconsistency of results where an exact link between alterations in the intestinal bacteria and the disease mechanisms cannot be made (18). Newberry and co-workers reported in their systematic literature review that there were eight agreeing and seven conflicting results between ME/CFS microbiome studies, while there is overall evidence for dysbiosis (19).

In one noteworthy study the investigators were able to classify 83% of patients correctly as ME/CFS patients or controls based on their evaluation of dysbiosis in gut microbiome and some increased inflammatory markers in blood as a consequence of microbial translocation, though sample size (N=48) was relatively small and future replication is needed. A reduction in bacterial diversity and richness, lower anti-inflammatory species carriage and higher levels of pro-inflammatory bacterial species such as Enterobacteriaceae were found in a 16S rRNA stool sample analysis (13). Normally, a high microbial diversity is associated with a healthy and fit human host. Cardiorespiratory fitness, which is measured by the maximum oxygen consumption (VO2 peak), maps to about 20% of the variation in gut diversity (159). This means that reduced gut gene richness and microbial diversity in ME/CFS patients also might be associated with a different physical fitness activity. Significantly lower VO2 peaks in ME/CFS patients have been described (162). While Armstrong and co-workers found significant absolute decreases of intestinal bacteria, mainly by the anaerobic bacteria (15), the opposite with an increase of anaerobic microorganisms has also been reported in ME/CFS patients (16).

Higher levels of Enterococcus spp (16), Streptococcus spp (16, 163), lower levels of the beneficial Bifidobacteria (13, 154) and decreased anti-inflammatory Firmicutes (13, 14) have been detected. Enterobacteriaceae were increased in several studies (13, 16, 22). According to the metagenomic analysis by Nagy-Szakal and co-workers, specific bacterial taxa such as the Firmicutes phylum, and the Faecalibacterium, Roseburia and Clostridium genera were associated with ME/CFS. In particular an increase of Alistipes and a decrease of butyrate-producer Faecalibacterium were considered as top biomarkers with potential diagnostic value (12). Their finding of a higher abundance of the Clostridium genus contradicts the opposite results reported by Giloteaux and co-workers (12, 13). The lower abundance of Faecalibacterium was also replicated (13). A case report published by the same author compared ME/CFS-discordant monozygotic twins and also identified abundance of Bifidobacterium and Faecalibacterium to be reduced in the affected twin compared to the non-affected one in addition to lower gut alpha diversity (164). Interestingly, a reduction of Faecalibacterium was also found in IBD patients with fatigue (compared to IBD patients without fatigue) (165), cancer-related fatigue (compared to cancer patients with low fatigue) (166) and other autoimmune diseases such as MS (167, 168) and diabetes type 1 (169).

Several studies reported a reduction of a butyrate-producing Bacteroides species in ME/CFS (19). High-throughput 16S rRNA sequencing from stool samples revealed gut microbiome alterations that were consistent with an increased inflammation of the gut, namely significantly increased Lactonifactor and Alistipes (14).

One study with 48 patients and 52 controls identified 26 markers including bacterial taxa that were distinct for ME/CFS and healthy controls, respectively (170). Gut abundances of Coprobacillus, of Eggerthella and Blautia were best able to distinguish patients from controls. Decreases in Faecalibacterium (12) and increases in Coprobacillus were found and also seen in other previous studies, as discussed above (13). To evaluate the diagnostic potential of the markers, THE researchers separated patients into short-term and long-term duration of ME/CFS. They found that microbiome changes were much more prevalent during the short disease duration (≤ 3 years) compared to a long disease duration (> 3 years). The authors suggest that these microbial markers mirror temporary changes in the early disease stage and hence show potential for future biomarker application (170). Furthermore, a study comparing the microbiome composition of ME/CFS patients to acute Q-fever patients suffering from fatigue as well as healthy controls, found the microbial signature of ME/CFS and Q-fever to be quite similar, while both groups differed profoundly from healthy controls. ME/CFS and Q-fever fatigue syndrome patients further shared the same set of pro-inflammatory markers with which a differentiation of each patient group from healthy controls was possible. The abundance of Firmicutes and Actinobacteria was increase in ME/CFS patients (n = 50) compared to healthy controls (n = 72) while Bacteroidetes were reduced (171).

Another recent study with 35 patients and 70 controls also found a distinct microbial pattern with decreased anti-inflammatory Firmicutes (117), confirming previous results (13, 14). This study compared patients with “internal” (relatives of the patients) vs. external controls and found several overlaps between patients with their relatives, but not with external controls, emphasizing the role of controlling for lifestyle, genetic confounders, and the choice of representative controls (117).

These partially inconsistent but evident findings of dysbiosis in ME/CFS have been described by previous reviews (18, 19), but their exact role in the disease mechanism remains unclear (Table 1). The above studies vary strongly with regards to sample size, recruitment (including severity at sampling), and methods for microbiota characterization (most studies using culturing or 16S amplicon sequencing, with only one whole genome shotgun sequencing study to date).

We should take into consideration that ME/CFS patients take many different drugs regularly to manage their variable symptoms (32, 54) and it is known that not only antibiotics, but also other drugs have a significant effect on the commensal microbiota (28, 172, 173), which future work must account for. Diet may also rapidly influence the microbiome in the short term (174). Long-term dietary habits have also been associated with distinctive faecal microbiome genera, e.g., high protein/animal fat diets are associated with Bacteroides, whereas diets high in carbohydrates are associated with a high prevalence of Prevotella (175). To account for this, dietary intake from at least 48-24 h prior to stool sample collection should be considered. In addition, larger studies with clear clinical inclusion criteria are needed to consider the different subgroups of patients with ME/CFS, as well as consider the viral and eukaryotic components of the gut microbiome (17–19, 176).

The old sayings to listen to one’s gut feelings, or that something upsets one’s stomach, may not have gained traction by coincidence. Several animal studies provide direct evidence that the intestinal microbiome communicates bidirectionally with the brain (177–179). Signals from the brain have consequences on our intestinal function and vice versa visceral afferents send signals to the brain. These effects can already be seen in daily life, where anxiety, emotional stress or pain influence stool regulation and can lead to diarrhea or obstipation. The routes of communication from the gut to the central nervous system are expected to involve at least 4 pathways, the (I) immune system pathway by cytokines, (II) the hormonal route by the neurotransmitters GABA, serotonin or dopamine, (III) neural routes including the vagus nerve and spinal pathway and (IV) the metabolic pathway including SCFA (180, 181). Neurotransmitters like dopamine or GABA and also signaling molecules are stimulated or generated by intestinal bacteria. Metabolites produced by bacteria are for example SCFAs acetate, butyrate, and propionate (152), and the amino acid and precursor of the neurotransmitter serotonin, tryptophan (182). Some bacteria are also known to produce neurotoxic metabolites (153). The gut-brain communication has been supported by finding signatures of the gut microbiome of increased risk of developing diseases such as post-traumatic stress disorders (183). Here the transplantation of intestinal bacteria from depressive patients to rats induced a typical depressive behavior in the germ-free animals (184). Overall the microbes in the intestine are able to send signals to the brain and influence sleep, stress, memory, mood and cognition (153).

Support for a role of gut-brain communication in ME/CFS has been provided from different fields of studies. Antimicrobial and probiotic interventions in ME/CFS patients provide first evidence for gut-brain communication because of an improvement of different neurocognitive symptoms (163, 185, 186). The D-lactic acid-theory discussed below is also based on the microbiota-gut-brain connection. This hypothesis states that accumulation of D-lactic acid through bacterial fermentation leads to its increase in blood and then in brain regions, where it causes neurological symptoms like fatigue (16, 24). Furthermore the improvement of ME/CFS symptoms after rectal infusions of cultured bacteria (20) and after antibiotic treatment provides evidence for a role played by gut-brain interaction in ME/CFS patients (21). Also in other diseases like depression, treatment with antibiotics suggests a reduction of some symptoms, according to a recently published trial (187). In contrast it should also be noted that antibiotics may also contribute to dysbiosis and disease, especially when used over a long time (188, 189).

Under normal composition of the gut, bacterial translocation through the gut barrier should not be possible (5). However, there is evidence that intestinal epithelial barrier dysfunction with concomitantly increased permeability, so-called “leaky gut”, is present under inflammatory changes in many diseases such as IBD, IBS, liver diseases, diabetes mellitus and chronic heart failure. Interestingly, in a majority of these diseases there is also an association with gut dysbiosis (190). The subsequent risk for developing ME/CFS in patients with IBD, especially Crohn’s disease, is significantly higher compared to individuals without IBD (3.7% vs. 0.2%) (9). The pathogenesis of the relationship between IBS and ME/CFS remains unclear. One explanation might be that some gut disorders seen in ME/CFS patients such as a dysfunctional gut barrier result in an “irritable” bowel (191). There are also speculations about the common pathophysiology of bacterial translocation (9). The reverse angle was taken by Nagy-Szakal and co-workers who suggested that IBS symptoms might be a consequence of the pathophysiological predisposition or the psychological stress ME/CFS patients have to deal with. In their metagenomic study these authors showed that the signature of the gut microbiome in ME/CFS patients is not fully reducible to an IBS comorbidity. They pointed out that since both syndromes show similar symptoms, some, such as fatigue, might be mistaken for ME/CFS signals and some “ME/CFS bacterial taxa” shown to be typical in different microbiome studies might be linked rather to the IBS comorbidity than to ME/CFS (12). Another possible explanation is the “dysbiotic march hypothesis” from IBS to ME/CFS by Berstad and co-workers. They suggested a three-step progression where excessive use of antibiotics in childhood might lead to typical abdominal symptoms of IBS and, years later, to the development of chronic inflammatory diseases possibly including ME/CFS (192). A similar hypothesis was stated by Kuo et al., who showed a significantly increased risk in women of developing ME/CFS following Helicobacter pylori associated peptic ulcer disease in a recent retrospective cohort study (193).

Also several research teams reported an increased intestinal permeability in ME/CFS patients supporting the widespread hypothesis of bacterial translocation into the systemic circulation (6, 9, 23, 97), in order to explain the abnormally increased levels of IgA and IgM against LPS in peripheral blood of ME/CFS patients would not be possible (67% and 40% of ME/CFS patients showed increased IgA levels and IgM levels, respectively, compared to 0% in controls). In addition, IgA levels increased with severity of the illness (22). Furthermore, the elevated levels of bacterial components in plasma of ME/CFS patients indicate increased gut permeability. According to Giloteaux and co-workers it might be possible that increased endotoxin levels damage the epithelial barrier and infiltrate into the bloodstream, where it provokes an immune response and results in systemic inflammation. The reason behind higher levels of the microbial translocation marker LPS in circulation might be the higher intestinal abundance of Gram-negative bacteria as the bacterial endotoxin, which is found in their outer membrane, is associated with gut dysbiosis (5, 13). Nagy-Szakal and co-workers detected in addition elevated levels of plasma ceramide and assumed that this might be a hydrolysis product of increased LPS. The authors speculated that higher ceramides in turn might lead to a toxic reaction in different cell types, including gut epithelial cells, resulting in a dysfunction of the epithelial barrier and increased gut permeability, noticing that higher levels of ceramines have also been found in other chronic diseases such as IBS, diabetes, cardiomyopathy, and atherosclerosis (122).

Another reason for increased intestinal permeability might come from the increased by-products of the amino acid fermentation to produce SCFA (15), as discussed below.

Bacterial translocation could also be important concerning PEM. Shukla and co-workers examined this connection in a small study by observing phenotype characteristics, post-exertion malaise and microbiome changes 15 minutes, 48 and 72 hours after a maximal exercise test. Besides the expected large changes in pain, fatigue and confusion, interestingly also the relative abundance of the intestinal microbiota changed differentially than in the control group. Following exercise, the genus Clostridium was increased in the blood fifteen minutes later and abundances of Bacilli were significantly higher 48 hours later, which might be due to increased post-exertional bacterial translocation. The authors also pointed out that researchers should rather focus on the post-exertional changes in intestinal composition, than on the dysbiosis itself (23).

Reduced abundance of SCFA producing bacteria, especially those which produce butyrate like Faecalibacterium (12, 13) Roseburia (Firmicutes) (13) or some Bacteroides spp (15)., has been observed by different teams in ME/CFS patients. Serotonin, GABA and SCFAs are produced by bacterial fermentation and are mediators of gut-brain communication. SCFAs such as butyrate are considered the main energy source of colonocytes and are speculated to be a key regulator in the neuro-immunoendocrine system. They are associated with intestinal health by having anti-inflammatory effects and by enhancing intestinal epithelium barrier function (194).

A high-throughput sequencing study revealed that in physically fit humans, high fecal butyrate levels strongly correlated also with extent of physical fitness, therefore also with higher VO₂ peaks, and interestingly, also with increased gut microbial diversity. Especially higher levels of butyrate-producing organisms were associated with better fitness and higher VO₂ peaks correlated in the fit individuals with lower LPS biosynthesis. While higher LPS levels in blood normally lead to a strong inflammatory response, lower LPS levels induced by physical activity may explain the reduction of inflammation in fit individuals (159). If we connect these study results from the healthy population with the clinical picture of ME/CFS, a possible explanation for the decreased abundance of butyrate-producing bacteria might be that a reduction of SCFA results in PEM which in turn promotes inactivity, and inactivity in turn could subsequently lead to increased LPS in ME/CFS patients. The inverse scenario may also hold, such as physical inactivity generally follows from ME/CFS symptoms. And this inactivity might lead over time to a reduction of SCFA producing bacteria. A possible third option is that the depleted SCFAs are a secondary effect of the disease. SCFA producing bacteria might for example be repressed by higher levels of D-lactate producing bacteria. Higher levels of D-lactate producing bacteria have previously been reported in ME/CFS patients (16) and possible effects have been explored (25).

In contrast to this butyrate deficiency hypothesis, one study revealed an unexpected significant increase of fecal butyrate, isovalerate and valerate concentrations in ME/CFS samples. These increased fecal SCFA levels correlated with an increased bacterial fermentation. One suggestion by the authors about these altered metabolites was that it might originate from an elevated pH level in the gut or also from gut dysbiosis (15). This finding is also interesting, as SCFAs also can have neurotoxic effects under some circumstances (153). Another possible explanation for higher fecal levels of SCFAs may be an uptake deficiency. Therefore it should be considered that the fecal excretion of SCFA cannot be used as a full reflection of the actual concentration and production of SCFA in the intestine but should be complemented with serum levels (195). However, butyrate itself in serum has never been measured in any study to date. Caution is required in interpreting these results because multiple causal and indirect scenarios are possible. For example, ME/CFS patients are frequently medicated (54) and prevented from physical activity (50), and these two facts might indirectly drive their aberrant intestinal microbiota. Longitudinal studies over a long period, ideally including very early or even better pre-disease states, would be necessary to investigate causality, by assessing the correlation between butyrate levels or dysbiosis in general and the extent of symptoms. These should also account for medication intake and for extent of physical activity as potential confounding factors.

Lactic acid is one of the main by-products of bacterial fermentation, and generates stereoisomers L-lactic acid or D-lactic acid. D-lactic acid is known to cross the blood-brain barrier (196, 197). There were some indications that the clinical presentation of acute D-lactic acidosis is similar to the neurological symptoms of ME/CFS patients and an excess production and absorption of D-lactic acid in blood and brain was proposed as a possible mechanism in ME/CFS (16, 163). Sheedy and co-workers explained the mitochondrial and neurocognitive dysfunction by increased colonization of D-lactic acid producing intestinal Gram-positive bacteria in fecal samples (enterococci and streptococci). They hypothetically assumed that blooms of these bacteria would lead to lower intestinal pH-levels, resulting in a higher gut permeability, and besides the already discussed systemic inflammation, immune activation and oxidative stress might lead to a possible increase of D-lactic acid in blood (16). Based on these findings Wallis and co-workers investigated if a decrease in neuropsychological symptoms could be observed after targeting the overgrowth of the lactic acid producing Gram-positive Streptococcus genus with antibiotics and probiotics. By reporting no improvements in fatigue, they raised doubt about the D-lactate theory (163). Interestingly and a major limitation so far for this theory, blood lactate levels have not been measured in ME/CFS patients yet (25), whereas an increase in cerebrospinal fluid (198) and surprisingly a decrease of absolute fecal lactate levels has been found (15). It is important to note that fecal lactate levels also depend on uptake by the host and therefore do not necessarily reflect the actual intestinal lactate metabolism.

We should take into account that D-lactic acidosis is a very rare condition in general (197) whereas ME/CFS is a chronic disease in contrast to e.g. medication-induced episodes of acute lactic acidosis which in general do not cause long-term sequelae. There are also critical voices that see the oxidative stress and D-Lactic acidosis hypothesis as a new hype in the probiotic market. It was pointed out that there has never been proof of negative symptoms due to D-lactic acid producing organisms (196), with the exception being in patients with a short bowel (199). Apart from these patients, the intake of D-lactic acid producing bacteria has been proven safe for human health (196).

One of the newer research topics about the etiology of ME/CFS patients is an abnormal metabolism of tryptophan to kynurenine including its catalyst indoleamine-2, 3-dioxygenase (IDO) expressed in antigen-presenting cells (26, 27, 200, 201). The precise measurement of tryptophan and its metabolites is an interesting approach for potential biomarkers and therapies in current general research (202). Physiologically, the essential amino acid tryptophan is metabolized approximately 5% to serotonin and melatonin, and degraded approximately 95% by the enzyme IDO to kynurenine (182). IDO and the kynurenine pathway play important roles in the immune system as immunoregulators and immunosuppressors (203). IDO is induced by infection and inflammation through pro-inflammatory cytokines and mediators, mainly by IFN-y, but also IFN-α, TNF-α and LPS (204). Elevated IDO expressions seem to be an important hallmark of chronic viral infections such as HIV, hepatitis B and C by building an immune tolerant environment. Although elevated levels of IDO seem to correlate with the severity of diseases, the relevance of these findings remains unestablished (205). IDO activates regulatory T-cells and suppresses effector T-cell proliferation and function by the degradation of tryptophan to kynurenine (203), which enhances immunologic tolerance towards chronic infections.

A team around Robert Phair recently published an interesting hypothesis about an insufficient kynurenine production in ME/CFS patients, called the “metabolic trap hypothesis”, resulting in an abnormal steady-state of antigen-presenting cells, serotonergic neurons as well as serotonin- and melatonin producing cells. They demonstrated in a small patient population strong support for this theory. The hypothesis implies gene mutations in the isoforms of IDO, which would result in low kynurenine levels and a state with impaired CNS, gastrointestinal, immune and energy metabolism functions. Their examined ME/CFS patients had many more gene mutations in one of the isoforms of IDO, IDO-2, and also the number of mutations correlated with the severity of ME/CFS symptoms. The proposed mathematical model suggested that a lack of the enzyme activity of IDO-2 leads to a stop of the degradation of tryptophan to kynurenine and therefore tryptophan levels will rise in blood and in the cells, which the authors linked to the typical pathological steady-state of ME/CFS. Tryptophan levels in the ME/CFS cells actually were increased and also the expected lower kynurenine concentrations were confirmed. The authors concluded that if tryptophan is increased and the pathway to kynurenine blocked, activity of the serotonin and melatonin pathway will be disrupted. The clinical presentation would fit into this model because serotonin and melatonin dysregulation may both be involved in the symptomatology of ME/CFS. The consequence of the deficient kynurenine production would be among other things an insufficient synthesis of NAD+, which would affect the energy metabolism and potentially explain the hypometabolic phenotype of ME/CFS (27). Furthermore, while low levels have been shown to protect and repair the integrity and permeability injury of intestinal tight junction caused by LPS, elevated levels of tryptophan do not show these properties (206).

Under this “metabolic trap hypothesis”, possible immunological and metabolic mechanisms in ME/CFS are highlighted, with additional implications for the microbiome in ME/CFS patients. On one hand serotonin as a neurotransmitter in the CNS and in the gut wall is known to play a key role as a hormone in the gut-brain crosstalk (207). On the other hand tryptophan is not only metabolized through the endogenous serotonin, kynurenine and melatonin pathway, but also by the gut microbiota into indole derivatives (208). This means that changes in the function of IDO also have effects on the fermentation of tryptophan and the microbiome. A reduced activity of IDO is known to lead to an impaired intestinal mucosal barrier, increased endotoxin translocation and chronic inflammation (209), all things we have discussed to be present in the disease.

In contrast, there also exist thoughts in the opposite direction of the “metabolic trap hypothesis”, namely about an increased kynurenine production in ME/CFS patients, instead of a decrease. The depletion of tryptophan and generation of kynurenine through an increased IDO activity resulting from an activated immune response is expressed by the kynurenine and tryptophan (KYN/TRP) ratio, which is therefore a marker for the IDO activity and for cellular immune response in in vitro studies (210, 211). The fact that high IDO levels are associated with various chronic inflammatory settings (209), might support the thought of an increased kynurenine production. Also, the KYN/TRP-ratio and some of the kynurenine pathway metabolites such as the neuroactive and excitotoxic quinolinic acid have been observed to be higher in many diseases (212) and symptoms overlap with ME/CFS, such as hyperalgesia or higher sensitivity to pain, light, sound or touch, called central sensitization (26). Additionally, EBV infection, a common trigger of ME/CFS, induces IDO to degrade tryptophan to kynurenine (213). Along another line of reasoning, tryptophan metabolism has also been examined in IBD and IBS, which both have been associated with ME/CFS (32, 54). Changes in tryptophan metabolism have been shown to be involved in IBD, revealing lower serum tryptophan concentrations and bacterial tryptophan metabolites as well as increased kynurenine and higher IDO-1 activity (214). These findings suggest that tryptophan and its metabolites might have an important effect on the gut microbiome, the gut mucosal immune activation and its interaction with the host such as in the immune response (208, 214, 215). A recent interventional clinical study investigated such a high kynurenine hypothesis, and they revealed no direct alteration of KYN/TRP ratio in diagnosed ME/CFS patients. The researchers concluded that an abnormal immune activation and kynurenine metabolism might play an important role in the initiation of the active fatigue response of ME/CFS, but probably not later during the established syndrome (26).

It is well known that the postnatal and early life use of antibiotics disrupts the neonatal microbiome and is associated with a higher risk of several diseases such as asthma, IBD or type 2 diabetes (145). Also in adults there is, for example, an increased risk of 13% to develop diabetes after an antibiotic treatment (216), which supports that changes in the microbiome alter some metabolic conditions. According to a very large retrospective study, one treatment with antibiotics also increases the risk of developing IBD by 84% and with every following treatment there is a 6% higher risk. The researchers were also able to show that the earlier and more frequently antibiotics were taken, the higher was the risk of IBD (217). IBD is also often mentioned in connection with ME/CFS, although the correlation remains unclear (9, 13). Interestingly, the mode of delivery influences the amount of antibiotic resistance genes in the microbiome, with vaginally delivered infants appearing to be relatively protected against antibiotic resistance (150). Berstad and co-workers suggested with their three-set “dysbiotic march hypothesis” an explanation for the high comorbidity in ME/CFS patients with IBS, as described above (192).

After the discovery of penicillin, antibiotics massively intruded into medical practice in the 1940s and 1960s (218). According to a recent report, antibiotics disrupt the bacterial composition in the gut for up to 4 years and to some extent permanently (145, 219, 220).

Interestingly, the altered taxa by antibiotics (219) overlap with some altered taxa in ME/CFS patients shown by Nagy-Szakal and co-workers (12). To our knowledge, surprisingly, there is currently no study that focused on the antibiotic intake as a cause, trigger or pre-existing factor for developing ME/CFS. As described above, patients with ME/CFS tend to have previously suffered from frequent infections and there is plenty of evidence for an infectious trigger, proven by history, microbial or antigen detection (54, 71, 72, 92, 104). This in turn is likely linked to a higher antibiotic use in these patients as well. As Kronman and co-workers also noted, this does not necessarily mean that children with more frequent infections and antibiotic intake are more vulnerable to IBD or ME/CFS. However, a possible mechanism could be that antibiotic use changed the microbiome in such a way that their intestinal microbiota became more susceptible to some infectious or non-infectious triggers for ME/CFS, for example by impaired capacity to produce protective anti-inflammatory metabolites or increased potential for extraintestinal infiltration (217). More medications than only primary/so-called antibiotics may still possess antibiotic activity; a previous study with 156 tested non-antibiotic common drugs showed that as many as 78% inhibited bacterial growth (28). It is known that antibiotics provoke the risk of D-lactate toxicity. A potential reason for this might be that antibiotic exposure leads to an increase of D-lactate-producing bacteria (29). This hypothesis would be congruent with the hypothesis of a D-lactic acidosis in ME/CFS patients (15, 198) and higher abundance of D-lactic acid-producing bacteria (16).

Interestingly, it might be possible to link the antibiotic intake to the gender imbalance observed in ME/CFS patients. Women are diagnosed with ME/CFS up to four times as often than men (221). A metaanalysis and systematic review showed that women have in general a 27% higher antibiotic exposure than men (reflecting, for example, treatment of urinary tract infections). In the age group of 16 - 54 years the difference was even higher with 40% more prescribed antibiotics for females (222). Also, the mitochondrial and oxidative stress-hypothesis could be explained by previous antibiotic intake, as one study showed that antibiotics induced ROS and led to oxidative damage in human cell lines and mice. According to this study it might be possible that some antibiotics not only target bacterial components, but also mitochondrial components (30). In ME/CFS patients there is evidence for higher levels of ROS (115, 116, 121).

In contrast to this hypothesis, antibiotics have been discussed as possible treatment for ME/CFS, with the aim to reduce the overgrowth of some species (21, 163). Jackson and colleagues revealed in a pilot study sleep improvement after antibiotic treatment, mainly due to a decrease in lactic acid producing bacteria. They assumed that sleep quality might be improved due to a more equilibrated microbiome with reduced levels of Gram-positive bacteria, leading to a reduction of proinflammatory cytokines (21). With an antibiotic and probiotic treatment over 4 weeks Wallis and colleagues could show some benefits in neurological symptoms including sleep, but not fatigue itself (163).

Although multiple studies show evidence that inflammation in the gut as well as several diseases are associated with an altered gut-brain communication, evidence for a successful therapy in response to these complaints is rare (180). Two widely discussed therapies are probiotic interventions (95, 185, 186, 223, 224) and fecal microbiota transplantation (FMT) (225–228).