ME/CFS is a common disease of unknown etiology characterized by postexertional malaise (PEM; a type of fatigability), cognitive disturbance, unrefreshing sleep, autonomic nerve dysfunction, and a few characteristic comorbidities, see, e.g., Ref. (1–3). It often starts with an infection and has a strong tendency to remain a chronic condition.

ME/CFS diagnostic criteria have gradually become more stringent, see, e.g., Ref. (2–7). These are based on somatic, often self-reported symptoms (8, 9). Although often used interchangeably, studies using the “CDC,” (also called the “Fukuda”) criteria (5) mainly use the term “CFS,” while those using the “Canada” (3) or International consensus (2) criteria use the term “ME.” This creates an ambiguity, which may explain some contradicting results. There are so far no specific laboratory tests (10) for ME/CFS diagnosis. Recently, a committee recommended a new name for ME/CFS, systemic exhaustion intolerance disease (SEID) (11, 12) with diagnostic criteria that emphasize PEM as the central ME/CFS symptom (13). The disease entity ME/CFS is not uncontroversial. Like many times before in medical history, psychiatric and somatic explanations compete with each other. A recent critical review, which emphasized psychiatric aspects, stated that “there is no convincing pathogenesis model for CFS” (14). However, in this review, we forward that evidence for a somatic origin of the disease is accumulating.

From a research perspective it is important that patients are diagnosed using strict criteria. A thorough clinical examination is necessary. It does not matter how sophisticated the analyses are in a study if patient selection is ambiguous. In the case of “fatigue” it is important to distinguish ME/CFS fatigue from other types of fatigue, such as burnout syndrome and depression, see, e.g., Ref. (15). In ME/CFS, repetition of a physical or mental exertion can reveal objective evidence of fatigability. This exertion-elicited fatigue, PEM, is required for the diagnosis of ME/CFS using the Canadian criteria (3), the International consensus (2), and the SEID (12), but not using CDC (5) criteria. Although the term “ME/CFS,” encompassing both “ME” and “CFS,” has a built-in ambiguity it covers much of the current studies and is operationally judged as the best available concept. Fatigue similar to PEM also occurs in Sjögren’s syndrome (SS), primary biliary cholangitis (also named primary biliary cirrhosis) (PBC), and systemic lupus erythematosus (SLE). The relation of ME/CFS to the similar condition Gulf War Illness (GWI) is uncertain, see, e.g., Ref. (16). However, a recent study describes a laboratory-based distinction between the two illnesses (17).

Recent ME/CFS reports brought optimism (18). National Institutes of Health in the US announced that it will prioritize the disease. Cornerstones are studies on PEM (12, 19) and effects of immunosuppressive treatment (20–22) although not substantiated in a phase III trial.

There are several partly competing explanatory models for ME/CFS, for example; autoimmunity, chronic infection, energy metabolic defect, imbalance in autonomous nervous system and/or hormones, and psychosomatic dysfunction. In this laboratory-oriented review, we present an overview of recent findings and attempt to bring a substantial portion of ME/CFS symptoms and its disease history into one explanatory model. The model draws analogies from more established autoimmune diseases (even if much remains to be understood in these too) is based on clinical experience and on recent immunometabolic results. Clues for further research are given in Table 1 and as separate statements in the text.

We propose a pathogenetic model reminiscent of current thinking on the pathogenesis of autoimmunity.

A genetically predisposed person (A) is exposed to successive infections (B), e.g., in the gastrointestinal tract—manifested as dysbiosis or irritable bowel syndrome (IBS)—or in the airways, with microbes carrying epitopes mimicking human self-epitopes, or microbes which activate autoreactive B cells to produce the so-called natural antibodies with non-rearranged germ line immunoglobulin genes. Such autoreactive B cells may be deleted or persist in a state of anergy (C). A proportion of these B cells remain in spleen and lymph nodes as memory IgM⁺, IgA⁺, or IgG⁺ B cells (D). Individuals differ in time and extent of encounters with autoreactivity eliciting microbes. Some encounters are here postulated to give rise to autoantibodies (E) against key enzymes in energy metabolism hence causing a defective aerobic energy metabolism and PEM, the central symptom of ME/CFS, others to fibromyalgia (FM), yet others to postural orthostatic tachycardia syndrome (POTS) or other comorbidities. If the autoimmunization events are independent of each other they can occur in any order. If there are cooperativity effects they may follow a rather specific order (F). The upper case letters refer to stages in Figures 1 and 2.

Thus, the basic property of ME/CFS patients would be a defect in tolerance coupled with a chance exposure to microbes carrying relevant mimicking autoantigen epitopes.

The italicized text of Figure 1 shows a hypothetical explanation of the events behind ME/CFS. A known function of microbes in the gut is to train, from within, the immune system to recognize and react correctly to microbes (including bacteria and viruses), which come from the outside (25–29). The correct reaction includes, among other things, anergy and unresponsiveness to microbial antigens that cross-react with self-antigens. It is known that ME/CFS patients often have IBS (30–32). In this IBS there is also a modified gut flora (33, 34). A less symptomatic gut dysbiosis may also occur (33, 35). In addition, there is also occasional epithelial barrier leakage of gut microbes. It is reasonable to assume that the innate mucosal immunity defenses have been breached or that peripheral tolerance maintenance (training function) of the gut flora has been disturbed. Normally, the mucosal immune system must maintain tolerance to harmless foreign antigens including food and commensal microbes. Presentation of antigens at mucosae often leads to tolerance (36). When there is microleakage tolerance may not function properly leading to loss of checkpoints that normally prevent development of autoreactivity (37–42). The profile of B cell subpopulations is different in ME/CFS compared with controls (43). A factor behind that could be new memory B cells with autoreactivity, which normally would be sorted out, arising and persisting. When the body is exposed to a new infection, these B cells could produce antibodies which react both to microbe and autoantigen. Autoantibodies and T cells that recognize self-peptides can damage cells which carry autoantigens. This is the so-called mimicry (antigen similarity) theory behind autoimmune disease (44). Part of the explanation for ME/CFS would then be the disturbed gut flora and microleakage from the gut. At the left side of Figure 1 is written “Genetic predisposition.” This is compatible with the increased frequency of ME/CFS in certain families. Like for many other common diseases ME/CFS could depend on both inheritance and environment.

If this hypothesis is correct, a tendency for autoreactivity would arise gradually, via a changed gut flora and microleakage from the gut. After a decisive immunization event autoimmunity leading to ME/CFS would arise, as shown in Figure 1. The prerequisites for autoimmunity would arise gradually because B cells with a tendency for autoimmunity would arise after recurring microleakage across the mucosal barrier of the gastrointestinal tract inducing a state of chronic inflammation. The normal contact between gut microbes and immune system occurs at the gut/mucosa interphase. Central tolerance often develops by elimination of autoreactive B cells. However, a proportion of autoreactive B cells remain which are kept unresponsive (anergic). When there is microleakage, the mucosal barrier is bypassed and tolerance may not be maintained. Autoreactive B cells can then be activated and differentiate to autopathic B cells.

The frequency of IBS, alterations in microbiome and extent of microleakage should be further studied in ME/CFS. Attempts to find autoreactive B cells to find their origin and their evolution should be made. Maybe it is possible to trace how they evolved by systematic sequencing of their antigen-binding structures (paratopes and idiotypes), from germ line to anti-gut microbe to autoimmune clone?

There is evidence for a strong genetic component in some autoimmune diseases, such as complement component deficiencies in SLE which may lead to reduced self-antigen elimination. Likewise, in ME/CFS, autoimmune diseases, for example, thyroid disease (45), SS (46), and SLE (47), often occur among relatives and sometimes among the patients themselves.

Presence of an HLA association is a hallmark of many autoimmune diseases. It indicates an aberrant immune presentation to either cytotoxic T cells (HLA Class I) or T helper cells (HLA Class II) which predisposes for autoimmunity. One study found an overrepresentation of HLA Class II DQA1*01, with an odds ratio of 1.93 (48).

Specific cytokine gene polymorphisms were observed; an increase of one, for TNFα, and a decrease of one, for IFNγ, were found in CFS (49).

Recent genome-wide association studies showed an increased frequency in ME/CFS of single-nucleotide polymorphisms (SNPs), some isolated, some concentrated to three gene regions: microtubule associated protein 7, CCDC7 (coiled-coil domain containing 7) and a T-cell receptor alpha chain gene (50). The latter may confine an increased tendency to autoimmunity. The comorbidity with autoimmune disease or disease having an increased prevalence of autoantibodies, e.g., FM (51–57), IBS (58–63), POTS (64), and hypothyroidism (45, 51, 55, 65–67), also indicate a tendency for autoimmunity in ME/CFS patients (further detailed in the section on autoreactivie B cell clones and autoantibodies, including Table 4). SS (46) and SLE (47) often occur among relatives and sometimes among the patients themselves.

IgG3 and mannose binding lectin deficiency were more common among ME/CFS patients than in controls (102, 103). IgG subclass deficiency is more frequent in ME/CFS than in controls (104, 105). Such deficiencies could increase the risk of recurrent infections.

In a genetic study concentrating on hormone and hormone receptor genes, certain TRPM3 and CHRNA2 SNPs were found to be more common in ME/CFS (106–108).

There is abundant evidence for infection as a trigger of chronic fatigue in a more general sense (often manifested as fatigability) (68, 72–74, 77–79, 82, 84, 118–128) (Table 2). But negative evidence also exists (129, 130). Some of this evidence is inconclusive (131, 132). Whether all these instances of postinfectious fatigability have identical properties (e.g., Do they fulfill criteria for PEM?; For ME/CFS?; How chronic are they?; etc.) should be systematically investigated. These infections can be traced in the patient history, by direct detection of the microbe(s) (133), or by detection of antibodies to the microbe(s) (94, 119, 133–146), see, however, Ref. (147).

How often does it happen that spouses are afflicted? This would advocate a transmissible factor rather than inheritance.

Epstein–Barr virus (EBV) seems to be a frequent trigger of ME/CFS (also referred to as “postviral fatigue”). Glandular fever (127), also called IM, is most frequently caused by EBV. A reasonably specific laboratory test for IM (the “Mono” test) is based on heterophilic antibodies, which bind to carbohydrate antigens on non-human erythrocytes (148–150). If infectious triggers of ME/CFS are investigated, a positive Mono test provides an often recorded marker. Other infections are often not diagnosed as objectively. EBV belongs to the herpes virus family. It can infect and remain latent in B cells. At primary infection, EBV triggers massive activation of multiple B cell clones each secreting monoclonal antibodies which are coded by immunoglobulin heavy and light (IGH and IGL) chain genes with unique variable [immunoglobulin heavy chain variable (IGHV) and IGLV] genes, with the so-called complementarity-determining regions. The result is a polyclonal B cell stimulation, with massive release of natural antibodies, including autoreactive antibodies. Besides B cell growth stimulation, EBV stimulates production of EBI3, one of two chains of the tolerance-control heterodimer cytokines IL27 and IL35 (151). Thus, EBV is deeply influencing immune functions. EBV is also coding for antigens with highly repetitive structure (e.g., Gly–Ala–Gly–Ala repeats in EBNA1). This may be a source for antigenic mimicry and development of autoreactivity. Both the B cell growth stimulation and such antigenic mimicry make EBV a prime suspect of inducing autoreactivity. There is a correlation between occurrence of IM and the autoimmune diseases MS (152–154) and SLE (155–158). How EBV is involved, be it frequent reactivations of latent EBV or defects in the T cell and NK cell surveillance mechanisms against the virus, is not clear, but the presence of EBV, and the immune response to it, should be compared in ME/CFS, MS, and SLE, see Ref. (134).

Summarizing, EBV is especially interesting as a facilitator of autoreactivity. Some autoantibodies may have an origin in a mimicry between EBV antigen and self-antigens. EBV is a ubiquitous virus. EBV can stimulate thousands of B cells to produce thousands of different antibodies, each with its own unique antigen-binding site. It often occurs as an eliciting factor triggering ME/CFS, in this case referred to as “postviral fatigue.” As mentioned, it stimulates growth of a wide variety of B cells, and it has viral proteins that can give rise to autoantibodies (159, 160).

Transmissibility is a microbial property. Most ME/CFS cases are sporadic (118). However, there are a few recorded outbreaks, where healthy ME/CFS patient contacts developed symptoms of the disease (118, 161), forming ME/CFS outbreaks in ME/CFS, indicating a transmissible agent.

Many observations support that a condition similar to ME/CFS occurs in approximately 10% of those who had Q fever, an infection with the bacterium Coxiella burnetii (69, 70, 74) which often occurs in outbreaks. Q fever is unevenly spread throughout the world. ME/CFS is more widespread. Q fever is therefore unlikely to be a common cause of ME/CFS, globally.

A chronic postinfectious fatigue/fatigability reminiscent of PEM occurs after a number of life-threatening virus infections (Table 2). There is not much antigenic similarity between these infective agents. Many infections which fundamentally challenge or reorganize the immune system give rise to a persistent, perhaps autoimmune, malfunctioning state.

Note that the number of ME/CFS cases triggered by severe zoonotic infections such as Ebola must be very small, on a global scale. Besides IM, mild respiratory infections like those caused by mycoplasma (162–168) and chlamydia (164, 169, 170), or general infections due to parvovirus B19 (136, 171) and herpes 6 and 7 (133), have been mentioned, although the diagnostic evidence is not strong.

Although much remains unclear before the role of infection in autoimmunity is understood, there are diseases where a known antigenic challenge triggers autoimmune disease, an infection (172) or a vaccination. Surprisingly often it is the brain that is the target for this autoimmunity (Table 3).

In addition to commonly known microbes (virus, bacteria, and protozoa), a large number of new ones have been discovered during the last 5 years (171, 173–179). Many of them are viruses which do not cause any known disease. We need to keep an eye on these microbes. Maybe there are some among them which can precipitate ME/CFS?

In these examples of infection elicited autoimmunity, the microbial antigen mimics epitope(s) on human cells. Such microbial epitopes may either be small molecules, like lipids, or added to proteins posttranslationally (172), randomly similar sequences, repetitive sequence motifs, or highly conserved antigenic structures.

An example of the former mechanism (posttranslational antigenic modification) is PBC. The antibodies are directed against a small fatty acid molecule, lipoic acid, added posttranslationally to a protein in the pyruvate dehydrogenase (PDH) enzyme complex. PDH is part of the energy producing machinery at the surface of mitochondria (180, 181), and governs the transition from glycolysis (anaerobic energy metabolism) to the tricarboxylic acid cycle and respiratory chain (aerobic energy metabolism), occurring inside mitochondria. Lipoylation is a posttranslational modification, which also occurs in a few bacteria, such as Novosphingobium (182). Gut infection with Novosphingobium is a possible cause of PBC. The PBC patients have a PEM reminiscent of the PEM of ME/CFS. Likewise, periodontal infection with Aggregatibacter, which citrullinates its own as well as human proteins, may provide the final trigger for rheumatoid arthritis (183).

An example of the latter mechanism (conserved epitopes) is a family of highly conserved proteins, which are present in both humans and microbes, called “heat shock proteins” (HSP). Antibodies against HSPs occur in many often studied autoimmune diseases, for example, MS and SLE (94, 184–188). We found a higher frequency and levels of antibodies against a specific portion of HSP60 in ME/CFS patients (94). Even though HSP60 is a mitochondrial protein it is unknown if these antibodies can influence mitochondrial function.

How does autoreactivity develop? Much remains to be learned. The adaptive portion of the immune system (B and T cells) has a formidable task, to distinguish “self” from “non-self,” i.e., autoantigens from antigens of invading microbes. After an infection the immune response is initially relying on players of the innate immune system with natural antibodies, receptors for pathogen-associated molecular patterns, danger-associated molecular patterns, and DNA sensors for exogenous pathogens. However, within a few weeks the immune system acquires a higher precision with the developing adaptive immune B and T cells ensuring that only the targeted microbe is destroyed. The target selection (and tolerance development) may go wrong. Some microbial targets are very similar to self-molecules. This is the basis for the so-called molecular mimicry theory, although the evolutionary lines for microbes and humans diverged long ago. There are still some structures, such as HSP, which have hardly changed at all since then. An immune defense against them thus constitutes a risk of promoting an autoreactive response.

So-called “natural antibodies,” which occur in all persons and mostly are of IgM nature, often both poly- and autoreactive, are produced by a CD20+, CD27+, and CD43+ subset of B cells (189). One function of these natural antibodies is scavenging of dead/apoptotic, damaged and infected cells. Only sometimes do they result in disease. In a healthy person, B cells that can produce autoantibodies often rest in an “anergic” state and do not produce their potentially damaging antibodies. They can be activated by the so-called “cell danger” signals. B cells which produce natural IgM have regulatory functions (190). Such “innate” immune cells which are on the border of auto- and alloreactivity may be starting points for development of autoimmune disease.

Disturbance in the composition of the gut microbiome, dysbiosis, has been detected in several diseases (33–35, 86, 191–195). A major function of the microbiome probably is to train the immune system (e.g., T cells, B cells, and dendritic cells) with a large variety of antigens. Disturbance in it may lead to a defective immune repertoire and imbalance of tolerance induction (196). As the tools for studying microbiota gradually become more precise, the possibility of more or less specific changes in microbiota predisposing to autoreactivity is increasingly being addressed. This is the case for type 1 diabetes (27, 197, 198), multiple sclerosis (199), rheumatoid arthritis (200), SLE (201), Behcet’s syndrome (202), autoimmune gastritis (203), and ankylosing spondylitis (204). ME/CFS patients also seem to have aberrations in their gut microbiota (33, 192, 205, 206).

A symptomatic variant of gut dysbiosis, IBS (207), a common comorbidity in ME/CFS, may influence mucosal tolerance induction. Indeed, ME/CFS with IBS was suggested to be a distinct subset of ME/CFS (208).

It is conceivable that if the mucosal barrier also is broken by microleakage (28, 34, 37, 41, 192, 209–210), tolerance development may become impaired, facilitating development of autoreactivity (37–42, 211–213). Autoimmunity often seems to be a hit and run phenomenon. However, a chronic underlying infection cannot be excluded, also in the ME/CFS case (214, 215).

The hypothesis presented in Figure 2 is based on findings regarding the IGHV gene sequence VH4-34 (23, 24) in SLE. It elaborates the immunoevolutionary aspect of Figure 1. Its explanatory model is similar to current thinking regarding the pathogenesis of autoimmune disease. It tries to clarify the genesis of autoreactive B cell clones from germ line to pathogenicity. The original specificity conferred to a B cell and immunoglobulin by VH4-34 is anti-branched lactosamine containing carbohydrates. This then gradually mutates, probably due to exposure to epitopes from commensal gut bacteria. The original specificity exists in the beginning of a mutational walk in the Vh genetic maze, an example of epistasis, where one mutation facilitates other mutations during avidity maturation. Other unknown antigenic stimulations then give rise to various autoantibodies, some with anti-DNA specificity. We hypothesize that pathogenic autoantibodies in ME/CFS are created by a similar mechanism. Thus, we postulate that a genetically predisposed person gets a deranged gut microbiome which gives rise to apathogenic B cells with a weak autospecificity. They are not eliminated due to a defective tolerance induction. Finally, these clones are given an antigenic stimulation from an exogenous infection which yields pathogenic B cell clones. Thus, memory B cell clones with a paratope spectrum derived from germline and subsequent exposure to commensal microbes, e.g., in gut, may be an important intermediary step before development of outright autoimmunity. It should be possible to follow the path to autopathic clones by isolation and sequencing the variable immunoglobulin chains in B lymphocytes in ME/CFS, like what was done with VH4-34.

The objective measurement of energy metabolism by repeated cardiopulmonary exercise testing revealed a defective aerobic energy production in ME/CFS (19). This is manifested as an abnormal fatigability. Fatigability, which may or may not be related to the PEM of ME/CFS (252–255), occurs as a major symptom in some autoimmune (6, 184, 256–261), mitochondrial (262, 263) and infectious (264) diseases. It remains to be studied how unique the PEM of ME/CFS is.

Several observations indicate that the oxygen dependent (aerobic) energy metabolism is disturbed in ME/CFS (8, 19, 262, 271–273) (Figure 3; Table 6). That disturbance may be the reason for PEM. Mitochondria are the main producers of energy. They derive from α-proteobacteria which, over one billion years ago, were taken up into eukaryotic cells, see, e.g., Ref. (274). It is not unreasonable to guess that an immune defense against an infecting bacterium can cause collateral damage to mitochondria. While there must be protective mechanisms against this (e.g., tolerization), they may not always work.

A new dimension for understanding ME/CFS was added by recent publications (275, 279, 280, 282, 284). They revealed profound metabolic differences between ME/CFS patients and controls. Some of these changes may derive from an abnormal mitochondrial function in ME/CFS. Whether these abnormalities have an autoimmune origin is not known.

As shown in Table 7, much work remains before the autoimmune nature of ME/CFS can be considered established.

Even if ME/CFS is of autoimmune origin, is it the metabolic block (275) or the autoantibodies to hormone receptors (93) which are most important for pathogenesis?

The mechanism behind the flare after exercise (238) is obscure. Maybe a mitochondrial defect can lead to an increased activity in the innate immune network.

The disturbance in one-carbon metabolism (277, 280) may or may not be related to the disturbed transition between glycolysis and TCA cycle. It is indicative of a wider metabolic derangement than a block of PDH (275) would be expected to lead to. There are several papers on hormones (322), including glucocorticoids (257) and transient receptor potential channel hormones (222), and their receptors (106, 107, 109), in ME/CFS. It is conceivable that parts of the autonomic dysfunction can be explained in this way.

ME/CFS is a challenge for the patients, for medical research and ethics, for all of public health, and for society. The intensified hunt for scientific evidence explaining ME/CFS has large consequences for many thousands of people. Many of the published results need repetition. But as shown in this article the signs that autoimmunity and energy metabolic deficiency is involved in the disease have increased. A hypothetical but logical path, from gastrointestinal tract dysbiosis, to formation of pathogenic autoimmune B cells, to inhibition of energy production and deficient cognition, to flares of cytokine production, can be delineated. The natural history indicates that in many cases infections can elicit or worsen this autoimmunity.

The recently intensified research on ME/CFS yielded many biomarker candidates, as mentioned in this study. The main consequence of this work is that the proposition that there is no logical somatic explanatory model for ME/CFS (14) can be refuted. However, like for virtually all autoimmune diseases, the explanatory model has several tentative steps which need further exploration. The elucidation of the molecular nature of circulating metabolic inhibitors in ME/CFS (275) is a central question. If they turn out to be immunoglobulins, they may directly yield diagnostically useful biomarkers and an explanation of the mechanism underlying ME/CFS.

The risk of giving a hypothetical unifying explanation, as in this study, is that hypothesis can be perceived as fact, and that it influences the perception of the disease. But contacts with ME/CFS patients and those who care for them have convinced us that most can handle the uncertainty that hypotheses involve. Without hypotheses we cannot direct the acquisition of further knowledge of ME/CFS.

JB conceived of the paper and wrote most of it. C-GG added substantial parts especially regarding the clinical aspects. AE participated in the writing. She is writing a book on ME/CFS, her comprehensive knowledge was valuable. MR participated in the writing. He concentrated on checking references. AR contributed substantially, especially regarding the immunological aspects.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.