Authors: Huimin Xu, Qiaoqi Li, Yingzhe Luo, Hong Zhu
Categories: Review, Myalgic encephalomyelitis/chronic fatigue syndrome, Cognition, Microglia, Astrocyte, Neurovascular unit, Neurometabolism, Excitation–inhibition balance
Source: Journal of Translational Medicine
Authors: Huimin Xu, Qiaoqi Li, Yingzhe Luo, Hong Zhu
Cognitive dysfunction (“brain fog”), encompassing impairments in attention, processing speed, memory, and executive function, is a prevalent and disabling feature of myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS). While systemic immune abnormalities are well documented, the central nervous system processes linking peripheral immune disturbance to cognitive impairment remain poorly defined, limiting mechanism-based stratification and therapeutic development.
This review synthesizes evidence from neuroimaging, neuroimmunology, neurovascular biology, and cellular metabolism to propose a brain-centered, cell-resolved framework for ME/CFS-associated cognitive dysfunction. Distinct from prior neuroimmune or neurovascular syntheses that emphasize systemic inflammation or vascular dysfunction in isolation, this review adopts a brain-centered perspective. Existing findings are organized around interactions among brain-resident cellular populations within the neurovascular and synaptic microenvironment. Across studies, human evidence is largely derived from cross-sectional imaging, biomarker, and physiological assessments, supplemented by mechanistic insights from experimental models. The strength and consistency of evidence vary across proposed neurovascular and glial alterations are supported by relatively convergent imaging and biomarker data, whereas neuronal network imbalance, oligodendrocyte involvement, and extracellular vesicle–mediated signaling remain more heterogeneous and hypothesis-generating. Accordingly, microglial priming, astrocytic dysfunction, excitation–inhibition imbalance, and neurovascular unit alterations are interpreted as associative and context-dependent processes rather than established causal drivers of cognitive impairment.
Taken together, primarily cross-sectional human studies complemented by experimental data support a model in which cognitive dysfunction in ME/CFS reflects persistent but potentially modifiable neuroimmune and neurometabolic dysregulation, rather than fixed structural neurodegeneration. The translational value of this framework lies in its capacity to hierarchize mechanistic pathways by evidential strength and cellular context, thereby informing hypothesis-driven patient stratification and prioritization of glial-, vascular-, or metabolism-targeted interventions for future testing. Longitudinal and interventional studies will be essential to determine causal relationships and to evaluate whether targeting specific brain-resident cellular processes can meaningfully improve cognitive outcomes in ME/CFS.
Myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) is a chronic, disabling condition characterized by profound fatigue, post-exertional malaise, unrefreshing sleep, pain, and a constellation of neurocognitive symptoms often described as “brain fog” [1]. Cognitive impairment, involving difficulties with attention, memory, processing speed, and executive function, is reported a substantial proportion of patients, with prevalence estimates reaching up to 89% in selected cohorts assessed using self-report questionnaires or neuropsychological testing, and contributes substantially to the social and occupational burden of the illness [1]. However, reported prevalence varies widely depending on diagnostic criteria, cohort composition, and assessment methodology, underscoring the heterogeneity of cognitive involvement in ME/CFS. The mechanisms underlying these cognitive deficits remain poorly understood, but a growing body of evidence implicates chronic low-grade neuroinflammation, altered neurovascular regulation, and synaptic dysregulation within the central nervous system (CNS) [2–4]. In ME/CFS, persistent activation of the immune system (potentially triggered by infection or other stressors) is hypothesized to lead to elevated pro-inflammatory cytokines that can alter brain function, disturb neural cell homeostasis, and disrupt cerebral blood flow (CBF) [5]. Rather than constituting a unidirectional cascade, immune–brain interactions in ME/CFS are increasingly thought to involve bidirectional and context-dependent feedback loops, including autonomic, vascular, and neuroimmune signaling pathways. Consistent with systemic immune involvement, case–control studies report reduced NK cell cytotoxicity and single-cell profiling identifies immune cell–type–specific transcriptomic dysregulation (notably in monocyte subsets) in ME/CFS compared with matched healthy controls [6, 7]. These peripheral immune abnormalities provide a biologically plausible upstream context through which immune-derived signals may access or modulate central nervous system processes, via humoral, vascular, autonomic, or cellular communication pathways. How these peripheral immune alterations engage central brain-resident cells remains incompletely resolved, but such perturbations in the brain’s cellular milieu are thought to produce the subjective feelings of mental fogginess and the objectively measured cognitive slowing observed in patients [1, 8].
Notably, the CNS manifestations of ME/CFS occur in the absence of radiologically apparent focal lesions or classical hallmarks of neurodegeneration detectable by conventional imaging, such as widespread neuronal loss or gross cortical atrophy [9]. This absence should be interpreted cautiously, as current neuroimaging and neuropathological approaches may lack the spatial resolution or molecular sensitivity required to detect subtle cellular, synaptic, or metabolic alterations. This shift has redirected investigative efforts toward more nuanced pathophysiological mechanisms operating within brain-resident cell populations and their interactions. These cells, including microglia, astrocytes, oligodendrocytes, neurons, endothelial cells with associated pericytes, and those of the choroid plexus and meningeal compartments, collectively sustain the neurovascular and synaptic environment essential for normal cognitive function. Dysregulation in any of these cellular players could, in principle, contribute to cognitive dysfunction. Indeed, emerging evidence from neuroimaging and biomarker studies in ME/CFS points to abnormalities in multiple cell types. However, these findings derive largely from cross-sectional imaging and biomarker studies, with variable effect sizes and methodological heterogeneity across cohorts. In vivo brain Positron-emission tomography (PET) imaging has detected elevated glial activation in widespread regions (suggesting microglial or astrocytic activation) correlating with cognitive symptoms [2], although interpretation is constrained by ligand specificity, binding variability, and limited cellular resolution, and magnetic resonance spectroscopy (MRS) has revealed widespread increases in metabolites like choline, myo-inositol, and lactate in patients’ brains, consistent with glial activation and neuroinflammatory changes [4, 10]. These findings are not uniform across studies and are influenced by cohort heterogeneity and methodological differences. At the same time, functional MRI and single-photon emission computed tomography (SPECT) studies report CBF reductions and brainstem connectivity alterations in ME/CFS, implicating the neurovascular unit (NVU) and autonomic regulatory circuits in the cognitive deficits [3, 11, 12], though effect sizes are modest and cross-sectional designs limit causal inference.
Despite an increasing number of neuroimaging, biomarker, and experimental studies implicating neuroinflammation and vascular dysregulation, the specific brain-cellular mechanisms driving cognitive dysfunction in ME/CFS remain poorly defined [3, 13–16]. Most existing studies have focused on systemic immune alterations, whereas the specific contributions of brain-resident cells, including microglia, astrocytes, oligodendrocytes, endothelial cells, and neurons, have received relatively limited and insufficiently systematic investigation [17]. Moreover, prior reviews have tended to address neuroimmune, neurovascular, or metabolic abnormalities as largely separate domains, without explicitly examining how brain-resident cellular interactions may integrate these processes. Understanding how these cells interact within the neurovascular and synaptic milieu may reveal unifying mechanisms linking immune activation to altered cognition in ME/CFS. Therefore, this review aims to provide a brain-centered, cell-resolved synthesis of current evidence, delineating plausible brain-resident cellular pathways underlying cognitive impairment and explicitly evaluating the strength, consistency, and limitations of the available data, while highlighting key gaps that should guide future research. To support this aim, we conducted a semi-systematic, mechanism-oriented literature survey across PubMed, Web of Science, and Scopus, covering studies published between January 2000 and July 2024. Search terms combined “ME/CFS” or “chronic fatigue syndrome” with “cognitive impairment” or “brain fog” and mechanistic keywords related to brain-resident cells, neurovascular function, and metabolic or autonomic regulation. Peer-reviewed studies providing mechanistic insights into central nervous system involvement were prioritized, whereas studies lacking relevance to brain-resident cellular or vascular mechanisms were excluded. Literature screening and interpretation were performed by two authors, with any discrepancies resolved through discussion.
To facilitate interpretation across heterogeneous datasets and to avoid overstating convergence where evidence quality varies, the mechanistic sections below are structured using a simple, transparent evidence-grading framework. Specifically, findings are discussed according to three (i) ME/CFS-specific human evidence, including neuroimaging, physiological, cerebrospinal fluid, and blood-based studies conducted directly in ME/CFS cohorts; (ii) indirect or correlational proxies, such as peripheral biomarkers or non–cell-specific central nervous system measures; and (iii) cross-condition or experimental inference derived from animal models or related disorders, which is explicitly identified as hypothesis-generating.
Importantly, ME/CFS is a clinically heterogeneous condition, and mechanistic signals may vary depending on diagnostic criteria, illness duration and severity, and physiological context, including orthostatic intolerance or post-exertional symptom exacerbation. Accordingly, findings throughout this section are interpreted with the assumption that cohort composition and symptom state at assessment can influence observed effect sizes and apparent mechanisms, particularly for neuroimaging, vascular measures, and neuroimmune biomarkers. This framing provides a structured context for the cell-specific mechanisms discussed below.
Microglia, the resident immune cells of the brain, are central orchestrators of neuroinflammation and synaptic remodeling. In this subsection, we first summarize ME/CFS-specific human evidence, and then clearly label cross-condition or animal-model findings as hypothesis-generating inference. In the context of ME/CFS, microglial dysfunction or persistent activation (“priming”) is often hypothesized to underlie the chronic cognitive disturbances [18]. A seminal PET imaging study by Nakatomi et al. demonstrated elevated binding of the translocator protein (TSPO, a non-cell-specific marker of glial immune activation that can reflect microglial and astrocytic states) in multiple brain regions of ME/CFS patients (cingulate, hippocampus, thalamus, amygdala, midbrain, pons), with 45–199% higher signal than in healthy controls [2]. Notably, the degree of TSPO binding in several regions correlated positively with patients’ self-reported cognitive impairment scores, suggesting an association between TSPO binding and self-reported cognitive symptom severity of “brain fog” symptoms [2]. However, a more recent TSPO-PET study did not find significant differences between ME/CFS and controls, highlighting interpretative complexities because TSPO binding can reflect varied glial states and is influenced by genotype [19]. Importantly, common TSPO binding variants (e.g., rs6971) can classify individuals into high-, mixed-, and low-affinity binders, and this effect is particularly prominent for many second-generation TSPO ligands; thus, differences in genotype stratification/exclusion and ligand choice (first- vs second-generation tracers) may partially account for discrepant findings across studies. Despite these mixed findings, indirect biomarkers further support microglial involvement. Elevated myo-inositol and choline signals on magnetic resonance spectroscopy have been reported in some ME/CFS cohorts and may reflect alterations in neurometabolic or glial-associated processes; however, these metabolites are not cell-specific and are influenced by voxel composition, regional sampling, and acquisition parameters. In parallel, increased circulating inflammatory cytokines (e.g. IL-1β, IL-6, TNF-α) correlate with worse executive function and memory in patients, with multiple ME/CFS cohorts demonstrating that these cytokine signatures arise in the context of altered NK-cell function and pro-inflammatory monocyte activation relative to healthy controls, suggesting a plausible peripheral-to-central link to microglial priming [4, 5].
Once activated (or primed by a previous insult such as an infection), microglia can release pro-inflammatory mediators (IL-1β, TNF, prostaglandins, etc.), reactive oxygen/nitrogen species, and excitotoxic molecules that perturb neuronal function [20]. They can also phagocytose synaptic structures via complement-tagging pathways and thereby disrupt neural circuits—shown in development and disease, including early Alzheimer’s models in which complement (C1q/C3) and microglia mediate synapse loss [21, 22]. In related neuroinflammatory disorders and experimental models, excessive microglial activation has been linked to cognitive impairment via NLRP3-inflammasome signaling; however, direct evidence supporting this mechanism in ME/CFS is currently lacking, and its relevance should therefore be regarded as hypothesis-generating rather than established [23–25]. By analogy with findings from other inflammatory conditions and experimental models, one plausible hypothesis is that primed microglia in ME/CFS may respond to minor physiological or immune stressors with exaggerated inflammatory signaling; however, this proposed mechanism has not yet been directly demonstrated in ME/CFS and should be regarded as hypothesis-generating (Fig. 1). Conceptually, alterations in synaptic integrity or functional disconnection within networks supporting attention, working memory, and executive control have been proposed as a potential consequence of chronic neuroimmune activation; however, direct evidence for cumulative synaptic loss in ME/CFS is currently lacking, and this mechanism should be regarded as a hypothesis requiring longitudinal or interventional validation. Consistent with this, ME/CFS neuroimaging studies report reduced functional connectivity within frontoparietal/default-mode/salience and brainstem–cerebellar circuits [12, 26]. With respect to in-vivo glial immune activation, TSPO-PET findings in ME/CFS are one study reported higher regional TSPO binding with symptom correlations, whereas another did not replicate group differences. Given the limited cellular specificity of TSPO and sensitivity to rs6971 genotype and tracer choice, TSPO-PET should be interpreted as suggestive rather than definitive evidence for microglial activation in ME/CFS [2].Fig. 1Microglial priming, synaptic loss, and network dysfunction in ME/CFS. This schematic figure illustrates how chronically primed microglia may drive cognitive dysfunction. In ME/CFS, microglia (purple cells) are hypothesized to exist in a primed state after an initial insult (e.g., infection or trauma). Subsequent mild triggers cause exaggerated microglial activation with release of pro-inflammatory cytokines (red arrows) and other mediators. These factors induce local synaptic loss or dysfunction (pruning of dendritic spines and synapses indicated by X marks) especially in brain regions important for cognition such as the hippocampus and prefrontal cortex. The loss of synapses and altered neurotransmission lead to impaired network connectivity (dashed lines between neural circuits) and reduced signal integration across cognitive networks. Over time, this manifests as deficits in memory, attention, and executive function (“brain fog”). The figure highlights a feed-forward cycle wherein microglial cytokines further recruit immune molecules that reinforce microglial activation, thereby sustaining a chronic neuroinflammatory state that underlies cognitive impairment in ME/CFS
Microglia also regulate the excitatory/inhibitory (E/I) balance of neural circuits through synaptic pruning and secretion of trophic factors. Microglia-derived brain-derived neurotrophic factor (BDNF) can influence neuronal chloride transporters (e.g., down-regulating KCC2), shifting GABAergic signaling and network excitability [27, 28] (Fig. 2). Chronic neuroinflammatory conditions may induce microglia to engage in aberrant synaptic pruning, a mechanism implicated in developmental disorders and also proposed to contribute to “chemobrain” and aging-related cognitive decline [29–31] (Fig. 2). Though direct demonstration in ME/CFS is lacking, it is reasonable to propose that synaptic changes potentially involving microglial mechanisms may be relevant to memory encoding and multitasking difficulties; this remains a hypothesis that requires longitudinal and interventional validation. It is noteworthy that post-mortem CNS tissue studies in ME/CFS are extremely limited, so we must infer microglial pathology largely from imaging and animal models [32]. In animal paradigms relevant to ME/CFS, peripheral immune activation has been shown to prime microglia. Rodent models of repeated low-dose lipopolysaccharide (LPS) exposure develop a primed microglial phenotype that upon subsequent challenge produces exaggerated CNS cytokine responses and cognitive dysfunction analogous to sustained fatigue and memory impairment [33, 34]. Such findings support a two-hit model in experimental systems; whether this framework applies to ME/CFS remains to be tested in longitudinal or interventional human studies [35, 36].Fig. 2E/I imbalance in hippocampal–prefrontal circuits. This schematic depicts how an imbalance between excitation (red) and inhibition (blue) in interconnected hippocampal and prefrontal cortical networks can lead to cognitive dysfunction in ME/CFS. In healthy circuits (left side of figure), excitatory neurons (pyramidal cells) and inhibitory interneurons (e.g., parvalbumin-positive cells) maintain balanced activity, producing coordinated oscillations that support working memory, attention, and learning. Microglia (purple) and astrocytes (green) contribute to this balance by regulating synapse number and neurotransmitter levels. In ME/CFS (right side of figure), chronic neuroinflammation and glial activation shift this there may be excessive synaptic pruning or loss of inhibitory interneurons (fewer blue neurons and synapses shown), along with microglia-derived factors (like BDNF and cytokines) that enhance neuronal excitability (depicted by more intense red neuron firing). Additionally, downregulation of the KCC2 chloride exporter in neurons (in response to inflammation) can make GABAergic transmission less inhibitory, further tipping towards hyperexcitation. The result is network hyperexcitability and desynchronization – illustrated by erratic firing and impaired gamma oscillations in the hippocampus-prefrontal pathway. Functionally, this manifests as deficits in memory formation (hippocampal function) and reduced attentional control (prefrontal function), consistent with the clinical “brain fog” in ME/CFS
Astrocytes are fundamental for maintaining neuronal health, synaptic function, and the integrity of the blood–brain barrier (BBB) [37]. Evidence for astrocyte involvement in ME/CFS is currently dominated by indirect neurometabolic imaging and physiological studies, with mechanistic specificity largely inferred from experimental literature. In ME/CFS, astrocytic dysfunction or reactive changes (astrogliosis) may contribute to cognitive symptoms through multiple mechanisms, including disturbed neurotransmitter homeostasis, reduced metabolic support to neurons, and altered neurovascular coupling (NVC) [37]. Although no direct histopathological studies of astrocytes in ME/CFS exist, indirect markers from neuroimaging point to astrocyte activation. Elevated myo-inositol and choline on brain MRS have been reported in ME/CFS and are often interpreted as glial/astrocytic signals (mIns and tCho are often interpreted as reflecting glial-related and membrane-turnover processes, but they are not specific to a single cell type and depend on voxel composition, region selection, and acquisition parameters) [4]. Similarly, the increased choline signal in some ME/CFS studies may be consistent with altered membrane turnover; however, attributing this change to astrocytic remodeling remains inferential without cell-specific validation [4]. Notably, astrocytes are responsible for clearing neurotransmitters such as glutamate and for buffering extracellular potassium, and chronic low-grade inflammation may impair astrocytic glutamate uptake through transporters including EAAT1/GLAST and EAAT2/GLT-1, leading to disrupted network excitability [38]. In line with this, ME/CFS patients show abnormal glutamate-related signals on MRS in some regions (direction varies by field strength/ROI), including reports of elevated Glu/Glx in recent cohorts [39]. More broadly, neuroinflammation is known to reduce astrocytes’ capacity for glutamate uptake and alter transporter regulation, creating conditions for excitotoxic stress on neurons [40]. It should be noted that astrocyte involvement in ME/CFS is inferred primarily from metabolic imaging and cross-disease experimental evidence, and no direct histopathological or longitudinal confirmation is currently available.
A unique role of astrocytes in cognition is their provision of metabolic substrates to neurons. Through the astrocyte–neuron lactate shuttle (ANLS), active neurons receive lactate produced by astrocytic glycolysis, which is used to sustain synaptic activity and memory encoding. Blocking astrocytic glycogenolysis or neuronal lactate uptake (MCT2) impairs long-term potentiation and memory, whereas exogenous lactate rescues these deficits [41]. If astrocyte energy metabolism is compromised (for example, by mitochondrial dysfunction or reduced cerebral perfusion), neurons in memory-critical regions like the hippocampus may be vulnerable to energy limitations during high cognitive demand; however, current ME/CFS evidence is largely indirect and cross-sectional [41]. Intriguingly, studies have found increased ventricular cerebrospinal fluid (CSF) lactate in ME/CFS patients, correlating with the severity of mental fatigue [10]. This elevated brain lactate may reflect a shift toward glycolytic metabolism caused by chronic mild hypoxia or mitochondrial inefficiency within the CNS, and findings from orthostatic testing further show reduced CBF in ME/CFS, which supports the presence of an energy-supply limitation during physiological or cognitive stress [3, 10]. While higher lactate might reflect a compensatory astrocyte response, it could also signify that neurons are not efficiently utilizing the lactate for energy, thereby impairing cognitive function. Additionally, pro-inflammatory cytokines such as IL-1β and TNF can alter astrocyte physiology by inducing nitric oxide (NO) production, which inhibits components of the mitochondrial respiratory chain. This supports a plausible mechanistic framework linking peripheral immune activation with central energy stress and cognitive slowing, but causal directionality in ME/CFS remains to be tested [42].
When astrocytes undergo chronic activation or reactive transformation (astrogliosis), they can aberrantly release gliotransmitters such as ATP and D-serine, disrupting the fine-tuned regulation of synaptic activity. Sustained ATP release may be linked with increased adenosine signaling that suppresses neuronal firing through A1 receptors, which may be relevant to mental fatigue [43, 44]. At the same time, reactive astrocytes may lose their ability to provide balanced modulatory signals like D-serine, impairing N-methyl-D-aspartate (NMDA) receptor-dependent plasticity and cognitive performance [45]. In neuroinflammatory conditions, astrocytes can undergo reactive transformation with both detrimental and protective phenotypes. Some shift toward neurotoxic programs, such as releasing glutamate and prostaglandin/COX-2–related products, whereas others adopt neurotrophic and protective profiles [46]. The balance of these in ME/CFS is unknown. However, research in other contexts provides clues. Inhibiting inflammatory pathways in astrocytes has been shown to prevent cognitive deficits in a neuroautoimmune model. Astrocytic TNFR1 signaling is required for EAE-associated hippocampal synaptic alterations and contextual memory impairment, and blocking this pathway prevents the resulting cognitive deficit [47]. It is plausible that in ME/CFS, a milieu of persistent cytokines tilts astrocytes toward a phenotype that does not optimally support synapses.
Astrocytes constitute a crucial component of the NVU, with their perivascular endfeet enwrapping cerebral microvessels to regulate local blood flow in concert with neuronal activity. Through calcium-dependent signaling and the release of vasoactive mediators such as prostaglandins and NO, astrocytes play a central role in NVC, the physiological mechanism that matches regional CBF to local neuronal metabolic demand [48]. When astrocyte–vascular communication is disrupted, NVC efficiency can deteriorate, impairing oxygen and glucose delivery to active cortical regions [48]. This mechanism has particular relevance for ME/CFS, as functional MRI and arterial spin labeling (ASL) studies have revealed atypical patterns of CBF and reduced task-related perfusion responses [49, 50]. Astrocyte dysfunction, potentially induced by inflammatory cytokines, oxidative stress, or mitochondrial inefficiency, may underlie such neurovascular dysregulation by weakening astrocytic calcium signaling and reducing propagation of vasodilatory signals along the microvascular network [51]. Overall, available ME/CFS data are compatible with altered glial-related metabolism and neurovascular physiology, but the specific contribution of astrocytes remains a working hypothesis that requires cell-resolved and longitudinal validation.
Cognitive performance depends critically on the maintenance of a finely tuned E/I balance within neural circuits, particularly in regions such as the hippocampus, which supports memory encoding, and the prefrontal cortex, which governs attention, working memory, and executive control [47]. In ME/CFS, direct electrophysiological and synaptic-level evidence is limited; therefore, E/I concepts are discussed as circuit-level working models anchored to imaging, symptom physiology, and cross-condition neuroimmune literature. Disruption of this balance, whether driven by excessive excitation or inadequate inhibition, can impair the temporal precision of neuronal firing and ultimately lead to slowed information processing and working-memory deficits that present as cognitive inefficiency [52]. In ME/CFS, neurons are not thought to undergo major structural loss or classical neurodegenerative change, although persistent neuroinflammatory and metabolic stress may impair synaptic signaling and disrupt normal patterns of neuronal excitability [2]. Inflammatory mediators such as cytokines and prostaglandins can alter ion-channel kinetics, glutamate turnover, and GABAergic transmission, thereby shifting the E/I ratio and reducing cognitive resilience [53]. Multiple mechanisms likely converge to produce this imbalance in ME/CFS, including microglia- and astrocyte-mediated modulation of synaptic tone, altered glutamate–GABA cycling, and impaired neuronal energy metabolism that constrains inhibitory interneuron activity [27, 39, 54]. Collectively, these processes may destabilize circuit-level excitability and contribute to the characteristic cognitive slowing and attentional deficits observed in patients [8].
Neurotransmitter alterations. Chronic neuroinflammation can profoundly influence the balance of excitatory and inhibitory neurotransmission. Elevated glutamate or reduced GABA within specific neural circuits can drive hyperexcitability and impair cognitive processing [55]. In ME/CFS, proton MRS findings are mixed. A recent study reported higher glutamate (and N-acetyl-aspartate) in patients versus controls [39], whereas ultra-high-field work emphasizes regional heterogeneity of neurometabolites rather than a uniform global shift [56]. Thus, the absence of a consistent bulk abnormality does not exclude synapse- or region-specific dysregulation. On the mechanistic side, activated microglia can elevate extracellular glutamate and impair astrocytic uptake (EAAT1/EAAT2), promoting synaptic excitotoxicity and signal-to-noise degradation [40, 57]. In parallel, pro-inflammatory cytokines weaken inhibitory signaling. IL-1β has been shown to reduce GABAA receptor-mediated currents in hippocampal neurons and diminish inhibitory tone, a mechanism associated with seizure susceptibility and cognitive impairment in experimental models, and complementary evidence indicates that TNF-α also downregulates GABAA receptor-mediated inhibition [58, 59]. Imaging evidence in ME/CFS is directionally consistent with E/I disturbance at the systems level. A meta-analytic and systematic literature base reports abnormal activation/connectivity in insula and thalamus and disrupted limbic–cortical coupling in ME/CFS-fatigue cohorts and extends to altered subcortical/DMN and brainstem–cerebellar interactions in newer datasets and a transdiagnostic meta-analysis of fatigue networks [11, 60, 61]. Clinically, sensory hypersensitivity and insomnia reported by many patients align with a hyperexcitable network phenotype that could arise from these neurotransmitter- and cytokine-mediated shifts.
Network oscillations and microcircuit dynamics. Evidence from autism spectrum disorder and inflammatory brain models illustrates this principle [52, 62]; however, direct electrophysiological evidence supporting excitation–inhibition imbalance in ME/CFS remains limited. In autism spectrum disorder and inflammatory brain models, reductions in PV^+^ interneuron function and excessive microglia-mediated synaptic pruning contribute to network hyperexcitability and cognitive deficits [52, 62]. Furthermore, a specialized subtype of interneurons known as chandelier cells targets the axon initial segment (AIS) of pyramidal neurons, where they exert powerful inhibitory control over the initiation of action potentials [63]. Microglia are known to physically associate with the AIS and modulate its structural integrity and synaptic coverage [64]. Extending these concepts to ME/CFS as a working hypothesis, chronic low-grade inflammation could subtly impair interneuron inhibitory control or disturb microglia–AIS interactions, resulting in the disinhibition of pyramidal neurons within prefrontal and hippocampal circuits [65, 66]. Such microcircuit alterations could plausibly contribute to aberrant high-frequency oscillations or a degraded signal-to-noise ratio during cognitively demanding tasks. This mechanistic framework aligns with the cognitive slowing, attentional instability, and multitasking difficulty commonly reported in ME/CFS, features consistent with a state of cortical hyperexcitability and disrupted oscillatory synchronization [2, 67–69] (Fig. 2).
Kynurenine pathway and neurotransmitter receptors. Systemic inflammation in ME/CFS may perturb tryptophan metabolism along the kynurenine pathway, shifting the balance toward elevated quinolinic acid, which is a neurotoxic NMDA receptor agonist, and reduced kynurenic acid, which is an endogenous NMDA antagonist. This biochemical profile would bias cortical networks toward excessive excitation and impair cognitive efficiency [70]. Although evidence in ME/CFS is still emerging, multiple studies report abnormalities in tryptophan/kynurenine metabolites in blood or CSF, consistent with dysregulated glutamatergic signaling that could contribute to “brain fog” [71]. In parallel, autoimmune mechanisms may further modulate neuronal excitability. Autoantibodies against neuronal surface receptors, most notably muscarinic acetylcholine (M3 and M4) receptors and β-adrenergic (β1 and β2) receptors, have been identified in subsets of ME/CFS patients and may alter G-protein–coupled receptor signaling in a manner that affects arousal, attention, and synaptic transmission [72]. Together, disturbances in the kynurenine–glutamate axis and receptor-targeting autoimmunity provide convergent routes to cortical hyperexcitability and degraded signal-to-noise during cognitively demanding tasks, which aligns with the attentional instability and slowed information processing described in ME/CFS [73].
Connectivity and “hypofrontality”. Paradoxically, an E/I imbalance in the brain may manifest in two seemingly opposite ways, namely hyper-activation characterized by excessively noisy signaling and hypo-activation characterized by insufficient engagement of neural networks [74]. In patients with ME/CFS, some functional MRI studies report increased activation in mid-brain and brainstem arousal centres during cognitive challenge (which may reflect compensatory recruitment), while others document reduced activation in frontal cortical regions [75–77]. One study of intra-brainstem connectivity found deficits in connectivity between rostral medulla and cuneiform nucleus accompanied by enhanced coupling of hippocampal and brainstem nuclei during task performance in ME/CFS patients [76]. This pattern suggests that disrupted E/I balance in subcortical arousal systems (e.g., reticular formation, thalamus) may force patients to over-recruit these systems, whereas cortical circuits become under-engaged due to inefficiencies in signal processing. Supporting this interpretation, ME/CFS cohorts exhibit atypical increases in connectivity between brainstem and forebrain regions that correlate with fatigue severity during cognitive tasks, and reduced coherence in frontoparietal and default-mode networks at rest [60]. One plausible interpretation is that chronic low-grade inflammation triggers compensatory increases in cortical inhibition to protect against hyperexcitability, but this adaptation may lead to functional hypofrontality and reduced network engagement, impairing executive control and attentional stability [78, 79] (Fig. 2). In this framework, subcortical arousal centres are driven to maintain vigilance at elevated cost, while cortical cognitive networks fail to reach optimal recruitment, producing the characteristic cognitive slowing, attentional lapses and multitasking difficulty documented in ME/CFS.
In summary, neurons in ME/CFS appear to operate under opposing influences of chronic neuroimmune activation—cytokine-driven excitatory stress and compensatory inhibitory upregulation. This push-and-pull likely narrows the dynamic range of neural circuits, rendering them either easily fatigued or insufficiently responsive during cognitive demand, consistent with patients’ slowed processing and memory lapses. Similar E/I imbalances have been demonstrated in inflammatory and fatigue-related conditions using EEG and MEG approaches [60, 74]. For now, this remains a plausible mechanistic model grounded in the known effects of cytokines and glial mediators on synaptic transmission [79]. Interpretation of cognitive mechanisms in ME/CFS also requires explicit attention to confounders that can independently alter cognition and neuroimaging signals, including sleep disturbance, chronic pain, mood symptoms, medication effects (e.g., sedatives, stimulants, analgesics), and orthostatic intolerance. These factors can influence attention, processing speed, and perceived “brain fog,” and may also modulate physiological measures such as CBF and network activation. Accordingly, mechanistic claims below are interpreted as conditional on adequate control or stratification for these variables, and future studies should report and adjust for them systematically.
Oligodendrocytes, the myelinating glial cells of the CNS, are essential for rapid signal conduction and also provide metabolic support to axons. In ME/CFS, evidence for oligodendrocyte involvement is currently limited to diffusion-based imaging correlates and indirect clinical phenotypes, with little direct molecular or histopathological validation. Even subtle changes in myelin integrity or oligodendrocyte function can slow neural communication and impair cognitive processing speed [80]. ME/CFS is not classically considered a demyelinating condition (there is no clinical evidence of lesions as in multiple sclerosis), but emerging data suggest there may be microstructural white matter abnormalities and altered brain connectivity that could implicate oligodendrocyte involvement [11].
Advanced neuroimaging in ME/CFS has revealed differences in white-matter structure compared to healthy controls. Diffusion tensor imaging (DTI) studies have identified microstructural abnormalities in white-matter tracts, particularly within brainstem pathways and frontal white matter, and these alterations correlate with symptom severity [81]. One early DTI report observed reduced fractional anisotropy in the internal capsule and corona radiata of ME/CFS patients, implying subtle myelin or axonal irregularities [82]. Although ME/CFS is not classically a demyelinating disorder, these findings suggest that oligodendrocytes or myelin sheaths may be affected by the chronic immune or stress milieu of the condition. The cytokine TNF-α, which is frequently elevated in ME/CFS, exerts a dual influence on oligodendrocyte biology. Signalling through TNFR1 is generally associated with myelin damage and oligodendrocyte loss, whereas signalling through TNFR2 is linked to remyelination and the maturation of oligodendrocyte precursor cells [83]. Thus, in ME/CFS a predominance of pro-inflammatory signalling might tilt the balance toward subtle myelin injury, slowing neural conduction and impairing information processing speed.
Cognitive slowing in ME/CFS may stem in part from compromised myelin-axon integrity. Patients frequently report needing longer to think or find words, and neuropsychological tests show prolonged reaction times [84]. Demyelination, including patchy myelin thinning, increases temporal dispersion of neural impulses. Moreover, myelin and the oligodendrocytes that produce it also provide metabolic support to axons, as oligodendrocytes export lactate and other metabolites to sustain axonal health and conduction during prolonged neuronal activity [85]. If oligodendrocytes are compromised, possibly through oxidative stress or energy deprivation as suggested in ME/CFS, axons may fail to receive adequate metabolic support and become vulnerable to functional conduction block during high-demand conditions [80]. Such conduction inefficiency could manifest as mental fatigue, slowed processing, and short attention span characteristic of ME/CFS.
Comparative evidence from multiple sclerosis is informative for hypothesis generation; however, ME/CFS is not a demyelinating disease, and direct evidence of oligodendrocyte pathology in ME/CFS remains scarce [86]. Similarly, in normal aging, degradation of frontal lobe white matter is consistently associated with slowed cognitive flexibility and processing speed [87]. Accordingly, it has been hypothesized that in ME/CFS inflammatory or immune-mediated processes could subtly affect oligodendrocyte function and myelin could contribute to the “slowed thinking” phenotype. Supporting this view, structural imaging studies have linked reduced prefrontal white matter volume or integrity with orthostatic intolerance and fatigue severity in ME/CFS, raising the possibility that vascular-autonomic factors (such as reduced cerebral perfusion) may impair oligodendrocyte function in vulnerable white-matter “watershed” regions [11, 88, 89].
It should be noted that direct evidence of oligodendroglial pathology in ME/CFS remains scarce. No biopsy data or CSF measurements of myelin basic protein have been published to date. However, insights can be drawn from animal and cellular models of chronic immune activation. In experimental paradigms involving persistent viral infection or post-viral fatigue, microglia and astrocytes produce reactive oxygen and nitrogen species, which can damage nearby oligodendrocytes and disrupt myelin maintenance [90]. Chronic oxidative and inflammatory stress is also known to impair oligodendrocyte precursor cell differentiation and limit remyelination capacity [91]. Moreover, recent research shows that the pyroptosis cell-death pathway, triggered by inflammasome activation (e.g., NLRP3), can occur in oligodendrocytes under neuroinflammatory conditions, leading to demyelination and glial loss [92]. If similar low-level inflammasome activation were present in ME/CFS, it could subtly hinder oligodendrocyte renewal and long-term myelin stability. Extrapolating from these findings, ME/CFS may reflect a subtle “myelin-aging” phenotype in which persistent inflammatory stress accelerates oligodendrocyte functional decline similar to processes seen in physiological aging, thereby contributing to reduced processing speed and mental fatigue.
While oligodendrocytes have not traditionally been the focus of ME/CFS research, their possible involvement in cognitive dysfunction deserves consideration. Diffusion-MRI studies demonstrating microstructural white-matter abnormalities, particularly within frontal and brainstem pathways, are consistent with subtle myelin or axonal alterations in this disorder [81, 89]. Future investigations using more specific approaches, such as myelin water imaging or assays of myelin-related proteins in blood and CSF, may clarify whether a low-grade myelinopathy contributes to the cognitive phenotype of ME/CFS. If confirmed, therapeutic approaches that enhance oligodendrocyte resilience by promoting remyelination or protecting against cytokine-mediated toxicity, including modulation of TNF-α signaling through TNFR2, could represent a promising avenue to alleviate cognitive symptoms.
Cognition depends on adequate cerebral perfusion and the tightly regulated microenvironment maintained by the NVU, which consists of endothelial cells, pericytes, astrocyte endfeet, and neurons [93, 94]. In ME/CFS, the most consistent human evidence implicates cerebrovascular dysregulation, including reduced cerebral blood flow (CBF) and abnormal perfusion responses during orthostatic stress [95, 96]. These findings support impaired NVU function as a plausible contributor to cognitive inefficiency. By contrast, direct evidence for blood–brain barrier (BBB) disruption in ME/CFS remains limited, and BBB-related mechanisms are therefore discussed below as hypothesis-generating rather than established features of the disorder.
Adequate CBF and perfusion coupling are fundamental for sustaining cognitive function, and multiple studies indicate that these processes are compromised in ME/CFS. Early SPECT imaging revealed global brain hypoperfusion, particularly within the brainstem, in affected patients [97]. More recent ASL-MRI studies have confirmed that both global and regional CBF are significantly lower in ME/CFS compared with controls [98]. Such chronic reductions in perfusion limit oxygen and glucose delivery to active neurons, functionally resembling a mild ischemic state that may impair cognitive efficiency. Orthostatic intolerance appears to further worsen cerebral perfusion deficits, as head-up tilt studies show that individuals with ME/CFS demonstrate substantial reductions in cerebral blood flow even when blood pressure and heart rate remain within normal limits, suggesting impaired cerebrovascular regulation [3]. Moreover, brainstem gray-matter volume correlates with pulse pressure, suggesting that the integrity of autonomic centers is essential for maintaining cerebral perfusion. Mechanistically, dysfunction of the NVU could explain both baseline hypoperfusion and the blunted compensatory responses during cognitive demand. Normally, active neurons trigger astrocytic calcium signaling and endothelial NO release to increase local blood flow, but chronic inflammation and oxidative stress can reduce NO bioavailability and cause astrocytic end-foot swelling, thereby disrupting neuron–vascular communication [99, 100] (Fig. 3). Consistent with this, flow-mediated dilation in peripheral arteries is reduced in ME/CFS, reflecting systemic endothelial dysfunction that may extend to the cerebral circulation [95].Fig. 3Neurovascular unit breakdown and impaired neurocognitive perfusion. This schematic figure illustrates the impact of neurovascular unit (NVU) dysfunction on cognitive function in ME/CFS. In a healthy NVU (left panel), endothelial cells (orange) with intact tight junctions form a robust BBB, and astrocyte endfeet (green) and pericytes (brown) support blood vessel regulation. Neuronal activity leads to vasodilation (wide vessel lumen) ensuring adequate oxygen and glucose delivery for cognitive processes. In ME/CFS (right panel), chronic inflammation and oxidative stress lead to NVU endothelial tight junctions are weakened (gap between cells shown), increasing BBB permeability. Serum factors and immune cells (small red and green icons) infiltrate the brain, activating microglia (red) and astrocytes, which release MMPs and inflammatory factors that further disrupt the barrier. Pericytes may detach, and astrocyte endfeet become swollen and less able to modulate blood flow. Consequently, cerebral blood flow is reduced or mis-timed relative to neural activity (narrower vessel lumen, blunted dilation response). Neurons (blue) in cognitive circuits suffer from energy insufficiency and heightened exposure to toxic metabolites, resulting in impaired synaptic function. Clinically, this manifests as an inability to sustain attention or mental effort, especially under conditions of orthostatic stress or prolonged cognitive demand. The figure emphasizes the feed-forward BBB leakage fuels neuroinflammation, which in turn worsens NVU integrity, contributing to ongoing cognitive dysfunction in ME/CFS
The BBB integrity is essential for maintaining the brain’s immune privilege and homeostasis. The BBB, formed by tight junctions between endothelial cells and supported by pericytes and astrocytic endfeet, restricts the entry of peripheral toxins and immune mediators [101]. If this barrier is compromised in ME/CFS, even for short periods, circulating cytokines or immune cells may gain access to the CNS and be associated with neuroinflammatory changes that could plausibly contribute to cognitive symptoms (Fig. 3), although direct evidence for BBB leakage and its temporal relationship to cognition in ME/CFS remains limited. Direct in-vivo evidence for BBB disruption in ME/CFS remains limited and, to date, no published findings have demonstrated gadolinium leakage on MRI or altered CSF-to-serum albumin ratios. However, indirect indications exist, as several studies have reported elevated levels of CNS-derived proteins in patient plasma, which may suggest subtle barrier leakage [5, 95, 102]. Systemic inflammation is well known to weaken BBB integrity. In vivo and in vitro models demonstrate that cytokines such as TNF-α and IL-1β can downregulate endothelial tight-junction proteins, thereby increasing permeability [103–105]. Given the low-grade systemic inflammation consistently reported in ME/CFS, particularly during post-exertional malaise or symptom exacerbation, it is plausible that transient increases in BBB permeability may occur in response to inflammatory stress [5, 106]. Supporting this mechanistic link, Epstein–Barr virus (EBV) dUTPase, a viral enzyme implicated in post-infectious fatigue, has been shown to induce sickness behavior in mice while altering the expression of genes associated with BBB integrity and synaptic architecture [107]. Such findings raise the possibility that persistent viral or inflammatory stimuli in ME/CFS could weaken endothelial barrier function and disrupt the neurovascular microenvironment, thereby contributing to cognitive dysfunction.
Because direct evidence for BBB leakage in ME/CFS is currently lacking, the following consequences are discussed primarily to illustrate mechanistic plausibility based on other neurological disorders. The consequences of BBB breakdown are well established across neurological disorders; in ME/CFS, however, evidence for BBB disruption remains indirect and inferential.When the BBB becomes leaky, serum components such as albumin and fibrinogen can enter the brain parenchyma and activate microglia and astrocytes, initiating local inflammation [108]. In parallel, leukocytes may extravasate and release additional cytokines, amplifying neuroinflammatory signaling. In neurodegenerative diseases, the extent of BBB disruption correlates closely with cognitive decline, and leakage of blood-derived proteins has been linked to microglial activation, synaptic injury, and reduced neuronal plasticity [109]. Disrupted BBB integrity not only allows neurotoxic plasma proteins like fibrinogen to enter and damage synapses and oligodendrocytes but also impairs waste clearance, promoting accumulation of neurotoxic metabolites [110]. If ME/CFS involves even mild or intermittent BBB dysfunction, a hypothetical feed-forward process could be proposed in which peripheral immune mediators infiltrate the CNS and activate resident glia, and the activated glia may release proteolytic enzymes such as matrix metalloproteinase-9 (MMP-9) that further degrade tight junctions and exacerbate barrier permeability [111, 112] (Fig. 3). This proposed loop remains inferential in ME/CFS and requires longitudinal and mechanistic validation.
The concept of “cognitive energy” in ME/CFS can be understood as a failure of the NVU to adequately meet the metabolic demands of neural activity. Under normal conditions, challenging cognitive tasks elicit localized increases in blood flow and glucose delivery, which are reflected in the BOLD signal on functional MRI [113]. In contrast, ME/CFS patients often cannot sustain cognitive exertion, and neuroimaging studies have shown that they recruit additional cortical or subcortical regions to maintain performance, suggesting inefficient NVC and compensatory overactivation [11]. One plausible interpretation is that microvascular dysregulation forces the brain to engage larger neural territories to accomplish equivalent tasks, leading to accelerated cognitive fatigue. Recent whole-brain MRSI work reported elevated MRS-derived brain temperature in multiple regions (such as insula, thalamus, cerebellum) in ME/CFS, converging with lactate abnormalities and consistent with impaired neurovascular/thermal homeostasis [4]. In parallel, plasma proteomic analyses have revealed evidence of endothelial activation and dysregulated coagulation pathways, indicative of a pro-thrombotic and hypoperfusive microenvironment that could compromise cerebral microcirculation [114]. Together, these findings converge on a unifying model in which NVU inefficiency and endothelial stress constrain the brain’s ability to sustain metabolic support for cognition, providing a physiological substrate for the mental exhaustion characteristic of ME/CFS.
Altogether, current findings point to the vascular system as an important mediator of cognitive dysfunction in ME/CFS. Through their control of BBB integrity and cerebral perfusion, endothelial cells bridge peripheral immune signals and central neural metabolism. Impaired NVU function may thus heighten inflammatory exposure while reducing energy delivery for cognitive processes [96, 108].
Extracellular vesicles (EVs), including exosomes and microvesicles, are membrane-bound particles released by cells that carry proteins, lipids, and nucleic acids involved in intercellular communication. Increasing evidence indicates that EVs participate in brain–immune signaling by traversing physiological barriers and modulating distant cellular targets [115, 116]. In ME/CFS, EVs represent a mechanistically plausible pathway through which peripheral disturbances (such as systemic immune activation or gut dysbiosis) could influence central nervous system function and contribute to cognitive symptoms [117–119]. However, direct EV-focused evidence in ME/CFS remains limited. Most mechanistic inferences are derived from studies in related neuroinflammatory, infectious, or metabolic disorders and should therefore be interpreted as hypothesis-generating rather than disease-defining. Importantly, this conceptual framework is supported by consistent ME/CFS-specific findings of peripheral immune dysregulation, including reduced natural killer cell cytotoxicity and immune cell–type–specific transcriptional alterations in peripheral blood mononuclear cells [6, 7], which provide a biologically relevant context in which EV-mediated immune signaling could operate.
Evs released from peripheral tissues during systemic inflammation or metabolic stress can act as long-range messengers between the body and brain. Circulating EVs are capable of crossing a permeable BBB or being internalized by endothelial cells, thereby delivering their molecular cargo to neural and glial targets [120]. Experimental studies in other inflammatory conditions suggest that peripheral extracellular vesicles can carry immune-related cargo and influence central glial responses after interacting with endothelial barriers [116, 120]. In ME/CFS, however, such mechanisms remain inferential and are best regarded as hypothesis-generating frameworks rather than demonstrated pathogenic pathways. By analogy, in ME/CFS, persistent immune activity stemming from latent infection, microbiome imbalance, or autoimmunity could generate a chronic flux of pro-inflammatory EVs that enter or signal across the BBB. Even modest but sustained EV trafficking of this kind may keep microglia in a primed, pro-inflammatory state and subtly impair neuronal network stability, thereby contributing to the cognitive dysfunction characteristic of the disease. It is worth noting that EVs can encapsulate mitochondrial DNA (mtDNA) and other mitochondrial constituents that act as DAMPs, activating innate immune pathways in glia (e.g., TLR9/cGAS–STING) and promoting neuroinflammation [121]. Consistent with this mechanism, ME/CFS shows systemic mitochondrial dysfunction and oxidative stress, providing a plausible upstream source of mtDNA-rich EVs released from stressed peripheral cells [122]. In parallel, EVs from activated immune cells can carry processed IL-1β and related inflammatory cargo sufficient to modulate neural cells and prime microglia [123]. If peripheral EV release is elevated in ME/CFS, as suggested by reports of increased plasma EV abundance and altered EV-associated cytokine and protein networks including changes after exertion, then EV levels or specific EV cargos could potentially reflect cognitive symptom burden and serve as a measurable peripheral-to-brain signaling readout in plasma [119]. Importantly, these EV alterations are observed against a background of established immune cell abnormalities in ME/CFS cohorts compared with healthy controls, strengthening their relevance as biomarkers of immune dysregulation rather than nonspecific epiphenomena [67].
Neuronal and glial populations within the CNS also secrete EVs, constituting a bidirectional communication axis that complements classic cell-to-cell signalling. Neuron-derived EVs encapsulate synaptic proteins and regulatory RNAs that modulate target-cell synaptic plasticity, dendritic complexity and TrkB signalling [124]. At the same time, activated microglia release EVs enriched in pro-inflammatory cytokines (e.g., IL-1β, TNF-α) and/or complement proteins, which when internalised by neurons or astrocytes may impair cellular viability or precipitate apoptosis [123]. Moreover, in neurodegenerative models microglial EVs carrying C1q/C3 have been shown to tag synapses for elimination, thereby contributing to synaptic loss independent of direct microglial-synapse contact [125]. Although brain-derived EV profiling remains essentially unexplored in ME/CFS, a shift toward a pro-inflammatory or complement-rich EV cargo provides a mechanistically plausible pathway for synaptic dysfunction and cognitive impairment in this condition [126].
EVs offer a stable, cell-type–informative substrate for liquid biopsy of central processes in blood and CSF. Their lipid bilayer and selective cargo loading enable longitudinal sampling of proteins and RNAs reflective of neuronal and glial states [127]. In neurological cohorts, neuron- and astrocyte-derived EVs (NDEs/ADEs) have yielded candidate markers related to cognitive decline and injury trajectories, supporting their use as non-invasive indices of neuroinflammatory and synaptic pathology [128]. Within ME/CFS, plasma proteomics indicates endothelial activation with perturbations across coagulation and complement pathways, nominating vascular-immune axes relevant to cognitive symptoms [129]. Targeted analysis of the EV fraction rather than bulk plasma may improve detection sensitivity for relevant pathway signatures, as supported by recent ME/CFS studies showing exertion-responsive remodeling of the EV proteome that includes immune-metabolic modules [130]. Importantly, EV-based biomarker studies are sensitive to pre-analytic variability, including differences in EV isolation methods (e.g., ultracentrifugation, size-exclusion chromatography, precipitation-based approaches), normalization strategies, and cell-of-origin enrichment, which can substantially influence EV yield and cargo composition. Evidence from other neuroinflammatory disorders indicates that astrocyte-derived EV cargo (e.g., complement or inflammatory mediators) associates with cognitive status, reinforcing the plausibility that CNS-enriched EV signals could index neuroimmune contributions to cognitive dysfunction [131]. Accordingly, careful methodological standardization and reporting will be essential for interpreting and comparing EV-based findings in ME/CFS. Collectively, these data position neuron- or astrocyte-enriched EV cargo as tractable candidates to stage cognitive involvement and to monitor treatment response in ME/CFS, while mechanistically anchoring biomarker signals to neurovascular-immune biology [127].
Collectively, EVs represent a plausible but currently unproven route for peripheral-to-CNS signaling in ME/CFS, and their mechanistic role in cognitive symptoms remains to be established in ME/CFS-specific studies. By transferring cytokines, microRNAs, and other DAMPs, EVs can sustain bidirectional immune communication between peripheral tissues and the brain [116]. Although direct evidence in ME/CFS remains limited, observations from other inflammatory and infectious disorders support their role in propagating neuroimmune signaling [114]. Experimental inhibition of EV release, including approaches that block purinergic P2×7 receptor–dependent shedding, as well as strategies that neutralize pathogenic vesicles, has been shown to reduce neuroinflammatory responses in preclinical models [132]. Defining how these vesicle-mediated pathways translate peripheral stressors such as infection or exertional load into central “brain fog” could provide a novel framework for understanding and potentially mitigating cognitive dysfunction in ME/CFS.
Dysautonomia, which includes orthostatic intolerance, reduced heart rate variability, and altered baroreflex control, is one of the most consistent physiological abnormalities reported in ME/CFS [133]. Mechanistic interpretation should account for clinical heterogeneity, particularly orthostatic intolerance and symptom state at testing, which can substantially modify cognitive performance and CBF-related physiology. Brain regions that coordinate autonomic regulation, particularly the brainstem and hypothalamus, are tightly interconnected with immune signaling networks and cortical circuits governing arousal and attention [134]. Disruption within this brain–autonomic interface may therefore contribute to cognitive impairment in ME/CFS through both direct neural effects on vigilance and through altered neuroimmune communication along vagal and sympathetic pathways [135].
The brainstem and its ascending reticular activating system, together with thalamocortical projections, form the core arousal networks that sustain alertness and attentional tone. Neuroimaging studies indicate that ME/CFS involves both structural and functional alterations within these circuits, with reduced brainstem grey-matter volume and disturbed connectivity correlating with impaired baroreflex control and autonomic instability [11, 135]. Task-based fMRI demonstrates hyperactivation of key nuclei, including the midbrain reticular formation, locus coeruleus, and dorsal raphe, during cognitive effort, which is consistent with compensatory recruitment of arousal circuits under sustained cognitive load [11]. Moreover, PET imaging using TSPO-PET studies have reported altered TSPO binding in brainstem regions in some ME/CFS cohorts; however, TSPO lacks microglia-specificity and is influenced by genotype and tracer characteristics. Accordingly, these observations are best interpreted as indirect evidence consistent with altered glial immune activity in arousal-related nuclei, rather than as definitive proof of microglia-driven causality for attentional symptoms [2].
The brain–autonomic axis prominently incorporates afferent signalling via the vagus nerve, whereby sensory fibres relay immune and metabolic information from peripheral organs (including gut and liver) to the nucleus tractus solitarius (NTS) in the medulla, thereby providing a direct route for peripheral inflammation to induce central symptoms such as fatigue and malaise [136]. Experimental activation of vagal afferents suppresses systemic cytokine release and modulates central microglial activity, suggesting that persistent vagal signalling may maintain a chronic neuroimmune state [137]. On the efferent side, the cholinergic anti-inflammatory pathway is mediated by vagal output to peripheral organs such as the spleen and restrains cytokine production through acetylcholine acting on α7 nicotinic receptors, thereby supporting immune homeostasis [138]. In the context of ME/CFS, continual afferent stimulation (for example due to microbiome dysbiosis or increased gut permeability) combined with dysfunctional efferent vagal response could leave systemic inflammation unchecked and facilitate neuroinflammation and cognitive impairment [139].
It is notable that population-based studies have found that vagotomy (surgical transection of the vagus nerve) is associated with a lower risk of certain neurodegenerative diseases, suggesting that vagal pathways may participate in pathological propagation of disease processes [140]. In ME/CFS one may hypothesise a dysbalanced vagal axis with excessive afferent (pro-inflammatory) signalling and impaired efferent (anti-inflammatory) output. An initial infectious or immune insult may up-regulate afferent vagal pathways, and persistent systemic immune activation could subsequently desensitise or impair the feedback (efferent) limb of the vagus. Neuroimaging and functional studies confirm that immune challenges activate neurons in the NTS of the medulla, which receives vagal afferent input [141]. Although direct evidence of NTS neuronal hyperactivity or injury in ME/CFS is lacking, a testable working hypothesis is that altered signaling within vagal–NTS pathways and broader brainstem autonomic networks may contribute to dysregulated autonomic control and state-dependent reductions in cerebral perfusion, thereby worsening attention and cognitive efficiency during physiological stress [32, 96].
The hypothalamus serves as a key integrative hub linking autonomic regulation, endocrine responses (notably the hypothalamic-pituitary-adrenal, HPA, axis) and higher-order cognitive–emotional networks. In ME/CFS, findings from meta-analyses and review studies indicate reduced HPA-axis activity that presents as lower morning cortisol levels, attenuated cortisol awakening responses, and blunted stress reactivity, and this pattern may weaken central anti-inflammatory capacity and allow persistent neuroimmune activation [142]. Concurrently, limbic and interoceptive regions including the amygdala and insular cortex show altered structure (e.g., increased amygdala volume) and enhanced connectivity within the salience/interoception network in ME/CFS, suggesting over-engagement of internal monitoring systems at the expense of externally directed cognitive resources [143]. Moreover, inflammatory signals may access hypothalamic and circumventricular organ sites lacking a fully intact blood–brain barrier, thereby altering neurotransmitter systems (e.g., orexin, acetylcholine) that govern arousal, sleep–wake regulation and cognitive stability [144]. Taken together, these observations support a model in which hypothalamic–limbic dysregulation in ME/CFS contributes to both the perception of fatigue and the documented deficits in attention, memory and cognitive endurance.
Beyond brain-based regulatory loops, autonomic dysregulation in ME/CFS may exert direct effects on cognition via haemodynamic and neuroimmune pathways. Upright posture in ME/CFS is frequently associated with reduced CBF, and tilt-table studies have demonstrated that cognitive performance deteriorates in parallel with CBF decline during orthostatic stress [3, 145]. Heart-rate variability (HRV), a sensitive index of autonomic balance, is consistently reduced in ME/CFS and correlates with fatigue severity and cognitive disturbance, implicating parasympathetic insufficiency in the cognitive phenotype [146]. Moreover, increased levels of adrenergic and muscarinic receptor autoantibodies in ME/CFS immune cells suggest a chronically overactive sympathetic state and indicate the possibility of parallel alterations in central receptor systems, providing a plausible mechanistic link between autonomic overdrive, microglial activation, and cognitive impairment [147]. Collectively, these data support a model in which autonomic instability in ME/CFS contributes not only to fatigue and orthostatic intolerance but also to impaired attention, working memory and processing speed via compromised cerebral perfusion and neuroimmune mediation.
Taken together, current evidence supports a working and testable model in which cognitive dysfunction in ME/CFS arises from bidirectional coupling between arousal-related brainstem circuits, cerebrovascular regulation, and immune signaling within the central nervous system microenvironment. In this framework, persistent low-grade neuroimmune activation within brainstem and hypothalamic regions may alter autonomic output and cerebrovascular control, thereby constraining cerebral perfusion and metabolic support during cognitive demand. Conversely, autonomic instability and impaired perfusion may further sensitize brain-resident glial cells to peripheral and central inflammatory cues, reinforcing a state of network inefficiency rather than overt structural damage. Importantly, this model does not imply a fixed causal sequence but instead proposes a dynamic interaction among neural, vascular, and immune processes that can be empirically tested using longitudinal designs, orthostatic or immune challenge paradigms, and multimodal neuroimaging approaches. Framing brain–autonomic dysregulation in ME/CFS as a context-dependent and reversible systems-level process provides a conceptual basis for integrating heterogeneous findings, while maintaining focus on brain-resident cellular and microenvironmental mechanisms relevant to cognitive impairment [73, 146, 148].
Integrative working model and testable predictions. Building on this systems-level framework, a parsimonious model consistent with current ME/CFS data proposes that immune and metabolic stressors interact with cerebrovascular regulation to create a constrained neurovascular and neurometabolic environment. In this context, NVU inefficiency and reduced perfusion reserve may increase susceptibility to glial immune activation and impair astrocyte-mediated metabolic and neurotransmitter support, thereby reducing neuronal signal-to-noise and cognitive endurance during demand. This model makes falsifiable (i) cognitive worsening during orthostatic or post-exertional stress should co-occur with measurable reductions in CBF or vascular reactivity and state-dependent shifts in glial/immune proxies; (ii) subgroups with prominent orthostatic intolerance should show larger coupling between perfusion metrics and processing speed; and (iii) interventions that improve perfusion reserve or reduce inflammatory signaling should preferentially improve attention/processing speed endpoints in biomarker-defined subgroups.
Investigating the neurobiological mechanisms of ME/CFS poses distinctive methodological challenges. These challenges are particularly shaped by ME/CFS-specific features, including post-exertional malaise, orthostatic intolerance, and pronounced symptom fluctuation, which directly affect the timing, tolerability, and interpretability of cognitive testing and neuroimaging protocols. Current evidence is constrained by small sample sizes, heterogeneous diagnostic criteria, and variability in experimental paradigms, complicating interpretation and replication [9, 11]. This section outlines key methodological considerations, including interpretation of existing data, limitations of current research tools, and strategies to improve future neurobiological investigations.
The marked clinical heterogeneity of ME/CFS complicates mechanistic interpretation. Cognitive deficits vary widely across patients, ranging from marked impairments in memory and attention to more subtle difficulties with concentration, and current diagnostic frameworks including the Fukuda (1994), Canadian Consensus, and Institute of Medicine (IOM) criteria differ in the extent to which they require cognitive symptoms for case definition [149, 150]. Earlier cohorts often included individuals without prominent neurocognitive involvement, reducing the signal-to-noise ratio in neuropsychological and imaging studies [8]. Stratification by illness severity, mode of onset (post-infectious versus insidious), and presence of orthostatic intolerance—operationalized using standardized tilt or active stand testing, post-exertional challenge responses, and inflammatory biomarker profiles—may help delineate more homogeneous mechanistic subgroups [151]. Emerging work further suggests that biomarker-based clustering, which distinguishes inflammation-dominant from autonomic-dominant phenotypes, may help clarify underlying neuroimmune mechanisms and improve the reproducibility of research findings [73].
Most available evidence in ME/CFS is cross-sectional, which limits the ability to determine causal relationships and leaves uncertainty regarding whether microglial activation drives cognitive decline or instead reflects a downstream consequence [73]. Longitudinal within-subject study designs that incorporate exertional or orthostatic challenges alongside repeated cognitive, biomarker, and imaging assessments are essential to establish temporal relationships and strengthen mechanistic inference. Exercise-challenge paradigms, which transiently induce post-exertional malaise, have provided valuable insight by revealing concomitant declines in cognitive performance and elevations in inflammatory mediators after exertion [152–154]. Similarly, imaging conducted immediately after orthostatic stress has shown reductions in CBF that correlate with impaired attention and processing speed [96, 155, 156]. Such dynamic paradigms complement static baseline comparisons and represent a critical methodological advance toward disentangling the temporal and mechanistic links among immune activation, vascular dysregulation, and cognitive dysfunction in ME/CFS.
Each neuroimaging modality deployed in ME/CFS research brings distinct strengths, yet equally notable limitations. The PET using TSPO ligands is among the most widely used in vivo markers associated with neuroinflammatory processes; however, TSPO signal interpretation requires caution, as TSPO expression is not specific to microglia and can reflect activation of astrocytes, endothelial cells, or other glial populations under certain conditions [157]. In addition, different TSPO ligands vary in sensitivity to binding-status genotype and in signal-to-noise characteristics (e.g., first-generation tracers versus higher-affinity second-generation ligands), which should be accounted for when comparing studies reporting mixed TSPO-PET results [158]. At a minimum, future TSPO-PET studies in ME/CFS should report binding-status genotype, ligand generation, acquisition parameters, and analysis pipelines to enable valid cross-study interpretation and replication. Additionally, MRS provides metabolic information such as elevations of myo-inositol or lactate that may reflect glial activation, although it typically relies on large voxels and therefore cannot precisely localize regional metabolic variation, which limits its spatial specificity [10, 159]. Structural MRI studies, including voxel-based morphometry, have produced inconsistent findings in ME/CFS. Some cohorts report grey-matter loss while others do not, which is likely related to small sample sizes, heterogeneity in scanner platforms and acquisition sequences, and limited statistical power [11]. DTI may detect microstructural white-matter changes, but reproducibility remains weak across sites and protocols [160]. A promising advance is multi-site harmonised imaging protocols with larger sample sizes, which may enhance effect size detection and generalisability.
Peripheral blood biomarkers are frequently used in ME/CFS research as peripheral correlates of central immune activity. Elevated circulating cytokines, including IL-6 and TNF-α, have been associated with symptom severity and cognitive dysfunction, suggesting a partial reflection of neuroimmune activation in the periphery [5, 161]. Importantly, these peripheral inflammatory signals reflect underlying immune cell dysfunction in ME/CFS, including impaired NK cell cytotoxic responses and altered transcriptional states of circulating monocytes and T cells compared with healthy controls [6, 7, 67]. Nevertheless, the BBB restricts molecular exchange, and peripheral cytokine levels alone cannot accurately capture central neuroinflammation; accordingly, peripheral biomarkers should be interpreted as indirect correlates rather than direct surrogates of CNS-resident mechanisms. The CSF studies, although limited by their invasiveness and small sample sizes, have reported mild increases in protein concentration, altered cytokine patterns, and occasional oligoclonal bands, however, such findings are infrequent and inconsistent across cohorts, and no reproducible CSF signature has been established in ME/CFS [16, 162, 163]. A promising development is the characterization of neuron- and glia-derived EVs isolated from blood, which can cross the BBB and retain CNS-specific cargo. In Alzheimer’s disease and other neurodegenerative conditions, plasma neuronal exosomes have been shown to carry synaptic and inflammatory proteins that mirror central pathology [164, 165]. However, EV-based analyses are constrained by variability in isolation methods, cell-of-origin enrichment, and normalization strategies, which may substantially influence cargo composition and reproducibility. Similar analytical approaches could be applied to ME/CFS to identify neuron-derived exosomal cytokines, glial markers, or synaptic proteins as minimally invasive proxies for neuroimmune dysregulation.
The absence of a definitive animal model of ME/CFS presents a major translational challenge. Current experimental proxies, including viral infection models, chronic low-dose LPS exposure and repeated stress paradigms, each reproduce selected elements of the syndrome such as fatigue or microglial activation but do not recapitulate the full clinical phenotype [166]. Animal studies of chronic immune activation demonstrate that repetitive viral-mimic injections can induce microglial activation and impair memory tasks such as novel-object recognition [167]. However, rodents cannot self-report “brain fog” or reliably model the multi-system fatigue, post-exertional malaise and cognitive fluctuation seen in humans. Even so, these models remain valuable for mechanistic investigation, including approaches such as targeted disruption of microglial inflammasome pathways or administration of anti-TNF agents to assess cognitive resilience. Standardisation would greatly benefit the field, and a reproducible post-infectious rodent paradigm that combines an acute infection trigger, stress exposure and restricted activity may provide stronger construct and predictive validity for ME/CFS research.
Induced pluripotent stem cell (iPSC) technology enables the generation of patient-specific neural cell types, offering a platform for mechanistic investigation of neuroimmune processes. Human iPSC-derived microglia, astrocytes and neurons have been utilised in studies of neurodevelopmental and neurodegenerative disorders and display functional inflammatory responses in vitro [168, 169]. In the context of ME/CFS, patient-derived iPSCs could be differentiated into relevant glial and neuronal subtypes and then exposed to patient serum or cytokine challenges to assess cellular behaviour. Key limitations of iPSC-based approaches include incomplete cellular maturation, batch-related variability, and uncertainty regarding the extent to which in vitro inflammatory responses accurately recapitulate in vivo ME/CFS pathophysiology. This approach may identify intrinsic differences in inflammatory priming or responsiveness that contribute to the disorder’s neurobiological features. In parallel, organ-on-chip models of the BBB incorporating endothelial cells under shear flow together with astrocytes and pericytes permit direct assessment of barrier integrity when exposed to patient-derived plasma. Such models have shown that cytokines including IL-6 and TNF-α disrupt tight-junction organisation and increase barrier permeability in vitro [170, 171]. Application of these in vitro platforms to ME/CFS may provide mechanistic evidence of altered glial behaviour or BBB vulnerability, thereby bridging the gap between peripheral immune signals and central neurocognitive dysfunction.
The complexity of measuring cognitive dysfunction in ME/CFS is self-reported “brain fog” often diverges from standard neuropsychological test outcomes, in part because typical assessments capture isolated tasks under ideal conditions, whereas patients commonly struggle with sustained, multitasking demands. A systematic review observed that test selection, duration and fatigue effects contribute to inconsistent neurocognitive profiles in ME/CFS [8]. Many studies highlight the need to control for confounders such as pain, sleep disturbance and depression; meta-analytic data indicate that cognitive deficits persist in ME/CFS even after adjusting for mood symptoms [1, 8]. Additional critical confounders include medication use, autonomic state, recent exertional load, and circadian timing of assessment, all of which should be systematically measured or controlled to ensure valid interpretation of cognitive outcomes in ME/CFS.Modern approaches include ecological momentary assessment of cognition, which better reflects real-world variability and has demonstrated reliability in clinical populations [172]. Finally, incorporating patient-informed functional outcomes, including validated questionnaires that directly relate objective test results to everyday cognitive efficiency and real-world functional capacity, may improve alignment between laboratory findings and lived experience [173]. To improve consistency, future studies should prioritize core cognitive domains including processing speed, sustained attention, and working memory, with assessments timed relative to post-exertional malaise and orthostatic stress.
Statistical and technological considerations remain central to advancing the neurobiological study of ME/CFS. Many investigations rely on small, underpowered samples, increasing susceptibility to both type I and type II errors and limiting reproducibility [174]. Greater methodological rigor that incorporates preregistered hypotheses, open data sharing, and pooled analyses using meta-analytic frameworks would substantially improve the robustness, reproducibility, and comparability of findings across studies [175]. Technological innovation offers parallel opportunities. Multi-modal imaging approaches, such as simultaneous PET–MRI, can integrate metabolic and functional readouts to directly relate neuroinflammatory signals to network-level activity [176]. In addition, electroencephalography provides a cost-effective tool for examining excitation and inhibition balance through analyses of oscillatory activity and coherence, and quantitative EEG studies in fatigue-related disorders including ME/CFS have reported altered resting-state connectivity and event-related potential changes that align with subtle inefficiencies in cognitive processing [177]. Together, these strategies can strengthen mechanistic inference and bridge the gap between molecular and systems-level markers of cognitive dysfunction in ME/CFS.
In sum, although mounting evidence implicates brain-resident cells in the cognitive dysfunction of ME/CFS, refinement in methodology remains imperative to validate mechanistic claims. Addressing clinical heterogeneity through more rigorous study design, adopting advanced technologies such as patient-derived iPSC models and high-resolution imaging with improved reproducibility, and committing to preregistration, open data sharing and harmonised protocols will enhance reliability and translational value [169, 178, 179].
Recognition that brain-resident cells and neuroinflammatory processes likely contribute to cognitive dysfunction in ME/CFS provides a mechanistic rationale for targeted intervention [18, 73]. At present, management remains largely symptomatic, and no pharmacological treatment has consistently demonstrated disease-modifying efficacy in controlled trials [180, 181]. Emerging data on glial activation, endothelial dysfunction and immune-metabolic disturbance nevertheless suggest several mechanistically informed therapeutic directions, which are outlined in the following sections [95]. Importantly, the therapeutic strategies discussed below are presented in a graded, hypothesis-driven manner rather than as equivalent or ready-for-translation options, and are considered with respect to evidential strength, feasibility, and potential risk in ME/CFS populations.
If microglial over-activation contributes to cognitive impairment, therapies that modulate microglial reactivity offer promising avenues. The antibiotic Minocycline has been shown to inhibit microglial activation and improve cognitive performance in animal models of neuroinflammation [182, 183]. Likewise, inhibitors of the NLRP3 inflammasome, such as MCC950, have prevented cognitive deficits in preclinical models of Alzheimer’s disease and MS by reducing IL-1β and IL-18 release [184, 185]. Although none of these approaches have been tested in ME/CFS, the supporting evidence derives primarily from non-ME/CFS experimental systems, and the translational distance to ME/CFS—particularly with respect to chronicity, dosing, and patient vulnerability—remains substantial. Additional therapeutic candidates include low-dose naltrexone, which has been proposed to attenuate microglial activation, and short-term use of corticosteroids that can cross BBB. These agents differ markedly in their mechanisms of action, ability to penetrate the central nervous system, and safety profiles, and should not be considered interchangeable microglial-modulating therapies. These approaches warrant careful early-phase clinical testing, while fully acknowledging the limitations and risks associated with prolonged corticosteroid exposure [186]. In ME/CFS, such risks are particularly relevant given the potential to exacerbate immune suppression, metabolic stress, and post-exertional symptom worsening.
Astrocyte–neuron metabolic coupling represents a promising therapeutic target. Mitochondrial-supportive supplements such as Coenzyme Q10 (CoQ10) and NADH have shown clinical benefit in reducing fatigue in ME/CFS, and may also provide indirect cognitive improvement by enhancing neuronal energy availability and supporting metabolic resilience [187, 188]. However, evidence directly linking astrocyte–neuron metabolic modulation to cognitive outcomes in ME/CFS remains limited, and these strategies should be regarded as hypothesis-generating rather than disease-modifying. Targeting the ANLS through approaches such as calcium-channel modulation with agents including Verapamil remains a speculative strategy, although the underlying biological rationale is plausible. Cognitive enhancers such as Methylphenidate or Modafinil, though not disease-modifying, may transiently counteract glial-induced synaptic suppression by increasing cortical metabolism [189, 190]. Off-label use of Memantine (an NMDA receptor antagonist) or Amantadine to reduce astrocyte-driven glutamate toxicity has been trialed in small ME/CFS and related cohorts with mixed results [191–194]. Care is required, as stimulants and NMDA-modulating agents may exacerbate underlying metabolic fragility or autonomic instability in susceptible ME/CFS patients.
Restoring E/I balance represents another mechanistic therapeutic avenue in ME/CFS. If cortical hyperexcitability contributes to cognitive symptoms, particularly in patients with sensory hypersensitivity or migraine-like features, interventions that rebalance excitatory and inhibitory tone may offer symptomatic benefit [195, 196]. Benzodiazepines that potentiate GABAA receptor activity can provide short-term relief from agitation or insomnia, but they often induce cognitive dulling and are unsuitable for long-term use [197, 198]. A more mechanistic approach derives from autism research, where the loop diuretic bumetanide lowers intracellular chloride by inhibiting the NKCC1 cotransporter, thereby enhancing the inhibitory action of GABA. Randomized and open-label trials have shown that bumetanide can partially normalize neural excitability and improve cognitive and social outcomes, supporting its potential to restore cortical network balance [199–201]. Extrapolation to ME/CFS is constrained by the absence of ME/CFS-specific biomarkers of excitation–inhibition imbalance and by differing developmental and pathophysiological contexts. Extrapolating to ME/CFS, a cautiously designed pilot trial could evaluate whether bumetanide mitigates cognitive or sensory symptoms in patients showing evidence of network hyperexcitability, though careful monitoring is essential given its diuretic and orthostatic effects. In addition, anticonvulsants such as gabapentin or low-dose topiramate, which reduce neuronal hyperexcitability and are prescribed off-label for fibromyalgia and chronic pain–related “fibrofog”, may confer ancillary benefit, although their cognitive efficacy in ME/CFS remains unproven [202, 203]. Sedation and cognitive slowing associated with these agents may offset any potential cognitive benefit in ME/CFS.
Protecting and restoring myelin represents another plausible therapeutic strategy for alleviating cognitive slowing in ME/CFS. If oligodendrocyte dysfunction contributes to slowed neural conduction and cognitive impairment, therapeutic strategies that promote remyelination merit consideration. Vitamin B12 has been recommended empirically in ME/CFS on the rationale that it supports myelin integrity and mitigates oxidative stress; animal studies demonstrate that B12 enhances myelin regeneration and stabilises microtubules after traumatic brain injury, and human observational work links B12 status to white-matter and cognitive metrics [204, 205]. Another candidate, Clemastine, an antihistamine repurposed for remyelination, improved nerve conduction velocity and myelin water fraction in a phase II trial of MS, suggesting that promoting oligodendrocyte maturation may be feasible [206]. Nevertheless, direct evidence for demyelination or remyelination failure in ME/CFS is limited, and remyelination strategies should be interpreted cautiously in the absence of robust white-matter biomarkers. Given their favourable safety profiles, these agents could be explored in ME/CFS, ideally alongside interventions targeting oxidative damage (e.g., N-acetylcysteine) to protect oligodendrocytes and myelin [207].
Enhancing cerebral blood flow and neurovascular unit function represents a physiologically plausible approach to address cognitive symptoms in ME/CFS, although direct links to objective cognitive endpoints remain limited. Patients frequently show signs of orthostatic intolerance and cerebral hypoperfusion, suggesting that augmenting brain blood flow could mitigate “brain fog”. Pharmacologic agents such as midodrine, an α₁-adrenergic agonist, and fludrocortisone, a mineralocorticoid that expands plasma volume, are clinically used for orthostatic hypotension and have been reported to improve upright blood pressure and, in some cases, subjective cognitive clarity in ME/CFS and postural tachycardia syndrome [208]. Volume expansion using isotonic saline or oral electrolyte supplementation can also transiently enhance cerebral perfusion and relieve orthostatic cognitive worsening [209]. To directly target the NVU, vasodilatory agents such as cilostazol, a phosphodiesterase-3 inhibitor that improves cerebral microcirculation and endothelial nitric-oxide signaling, or acetazolamide, a carbonic-anhydrase inhibitor used in cerebrovascular reactivity disorders, could be explored. Cilostazol has demonstrated increased regional CBF and cognitive benefit in vascular and Alzheimer’s disease cohorts [210, 211]. Preserving endothelial integrity is another key component of NVU protection. When exercise is tolerated, even light aerobic activity can enhance endothelial nitric-oxide availability and improve vascular compliance; in more severe patients, neuromuscular electrical stimulation may mimic some of these hemodynamic effects [212, 213]. Dietary or pharmacologic antioxidants such as vitamins C and E and polyphenols including green-tea catechins may help counter oxidative stress that contributes to endothelial dysfunction, which in turn could support cerebral perfusion and promote cognitive resilience [214].
If BBB disruption contributes to ME/CFS pathophysiology, therapies that stabilise barrier integrity may be beneficial. The mast-cell stabiliser sodium cromoglycate can reduce histamine-induced BBB permeability, as mast-cell activation near cerebral vessels increases endothelial leakiness [215]. Some reports suggest that H1/H2 antihistamines or cromoglycate relieve “brain fog” in allergic or mast-cell–high ME/CFS subgroups, consistent with observations in mast cell activation and POTS/long-COVID cohorts where mast-cell–directed therapy is associated with improvements in cognitive and neuropsychiatric symptoms [216, 217]. The anti-inflammatory agent ibudilast, which inhibits glial activation and enhances BBB function, has shown efficacy in progressive multiple sclerosis and could be explored as a candidate for ME/CFS once safety and mechanistic data are available [218].
Targeting extracellular vesicles and their upstream signalling pathways represents an emerging but highly experimental mechanistic concept rather than a near-term therapeutic strategy for cognitive dysfunction in ME/CFS. Activation of the purinergic P2×7 receptor on immune and glial cells has been shown to drive extracellular vesicle release and facilitate the transfer of inflammatory mediators such as TNF-α, IL-1β and regulatory microRNAs across physiological compartments and into the CNS [132, 219]. Pharmacological inhibition of P2×7 R (for example with Brilliant Blue G) has mitigated neuroinflammation and cognitive impairment in animal models, suggesting that a similar strategy might attenuate EV-mediated brain signalling in ME/CFS [220]. Similarly, neutralising pathogenic extracellular vesicle cargo, for instance by using anti-TNF biologics to reduce or remove vesicles enriched with TNF, may provide a strategy to protect the brain from peripheral immune-derived insults. While systemic immunosuppressants (for example Rituximab) have been evaluated in randomized controlled trials in ME/CFS, these studies have yielded largely negative or inconclusive outcomes, these observations reinforce the need for targeted, barrier-safe interventions rather than broad immunosuppression [221].
Given the pivotal role of the vagus nerve in inflammation and brain–body communication, therapies such as vagus nerve stimulation (VNS) merit focused consideration. Non-invasive transcutaneous auricular VNS (taVNS) devices stimulate the auricular branch of the vagus nerve and thereby engage central vagal pathways without surgical implantation. taVNS has demonstrated anti-inflammatory effects and improved autonomic balance in diverse conditions marked by autonomic and inflammatory dysregulation [222]. In early pilot studies involving patients with fibromyalgia, a condition that shares substantial clinical features with ME/CFS, both taVNS and HRV biofeedback have been associated with improvements in autonomic function, pain symptoms, and subjective cognitive clarity [223]. Regular sessions of taVNS might therefore shift autonomic balance toward parasympathetic dominance, activate the cholinergic anti-inflammatory pathway, improve cerebral perfusion, and indirectly reduce microglial activation. The breathing exercises or HRV biofeedback that enhance vagal tone constitute low-risk adjuncts to modulate the autonomic–immune interface and potentially bolster cognitive resilience in ME/CFS. Regular sessions of taVNS might therefore shift autonomic balance toward parasympathetic dominance, activate the cholinergic anti-inflammatory pathway, improve cerebral perfusion, and indirectly reduce microglial activation. Complementarily, breathing exercises or HRV biofeedback that enhance vagal tone constitute low-risk adjuncts to modulate the autonomic–immune interface and potentially bolster cognitive resilience in ME/CFS [224, 225].
Symptomatic cognitive enhancers, as discussed above, may offer adjunctive support for selected patients with ME/CFS. Wakefulness-promoting agents such as Modafinil have been shown to improve attention and executive processing in controlled studies, although the effects are modest and not consistently replicated [226]. Ampakines, which act as positive allosteric modulators of AMPA receptors, have been shown to enhance synaptic transmission and cognitive performance in preclinical models, although evidence in humans remains limited [227]. Cholinergic strategies, including nicotinic agonists or cholinesterase inhibitors such as Donepezil, have been explored in small, off-label applications; transient improvements in mental clarity have been reported anecdotally, yet robust clinical validation is lacking [228, 229]. None of these approaches target the underlying pathophysiology, and their benefits often diminish over time, underscoring the importance of integrating them with disease-modifying or neuroinflammatory-targeted therapies.
In proposing these therapeutic hypotheses, it is important to emphasise that they remain largely unproven in ME/CFS and require rigorous clinical trials to assess both efficacy and safety [230]. Both mechanistic biomarker studies and subgroup analyses indicate substantial heterogeneity in ME/CFS, with distinct immunological, metabolic, autonomic and neurovascular profiles becoming evident across different patient groups [231, 232]. A matrix approach that matches patient subgroups (e.g., inflammation-dominant vs. autonomic-dominant) with targeted interventions, using measurable biomarkers and cognitive domain outcomes, may enhance trial design and therapeutic precision [233]. Importantly, given the chronic nature of ME/CFS, long-term safety, tolerability, and the risk of symptom exacerbation with prolonged exposure must be carefully weighed for any proposed intervention, particularly those affecting immune, metabolic, or autonomic function. Accordingly, we propose a biomarker-guided trial-design framework in which patients are prospectively stratified according to dominant biological profiles (e.g., inflammation-dominant, autonomic-dominant, or perfusion-limited) and matched to mechanistically aligned interventions, with predefined cognitive domain endpoints and safety monitoring.
Cognitive dysfunction in ME/CFS, commonly described as “brain fog”, has become a central focus of mechanistic investigation. The available evidence indicates that the cognitive impairment observed in ME/CFS is not due to a single focal lesion or neurodegenerative process but arises from a constellation of subtle, interacting abnormalities in brain-resident cells and their microenvironment. Microglia, the brain’s innate immune sentinels, appear capable of assuming a chronically primed state that perturbs synaptic function and network activity even in the absence of overt structural pathology [2]. Astrocytes may contribute through impaired neurotransmitter recycling and reduced metabolic support to neurons, consistent with altered glial metabolites and NVC reported in ME/CFS cohorts [234]. Neurons themselves seem structurally preserved yet functionally strained by excitation–inhibition imbalance and energy insufficiency, which may explain the characteristic slowing of cognitive processing [235]. Subtle involvement of oligodendrocytes and myelin integrity has also been proposed, aligning with clinical observations of reduced information-processing speed [236]. At a systems level, the NVU, which normally coordinates CBF, barrier integrity and metabolic exchange, appears compromised in ME/CFS and may contribute to reduced cerebral perfusion as well as increased BBB permeability, creating conditions that can further amplify neuroinflammation [96, 145]. Taken together, current evidence most consistently supports neurovascular dysfunction and glial activation as primary, imaging- and biomarker-supported mechanisms, whereas neuronal network imbalance, oligodendrocyte involvement, and extracellular-vesicle–mediated signaling remain secondary or hypothesis-generating.
Operationally, mechanisms can be provisionally ranked from those supported by direct human neuroimaging and biomarker data (e.g., cerebral hypoperfusion, glial activation) to those inferred primarily from cross-condition analogy or experimental models (e.g., myelin remodeling and extracellular-vesicle signaling). The strength of evidence supporting these mechanisms varies. Microglial activation and impaired CBF are directly supported by neuroimaging and biomarker studies, whereas oligodendrocyte pathology and extracellular-vesicle–mediated signaling remain speculative, inferred from analogous conditions or preliminary findings [2, 145, 236]. Nevertheless, the available evidence is most appropriately interpreted as supporting a working model in which ME/CFS reflects a state of chronic, low-grade neuroimmune and neurometabolic dysfunction, rather than definitive proof of reversibility or causality [235]. This framework provides cautious optimism because if neurons remain largely intact, cognitive deficits may improve once the surrounding cellular environment is normalized. This interpretation does not exclude psychological or behavioral modifiers of symptom experience but emphasizes that measurable neurobiological processes provide a necessary substrate for cognitive impairment in ME/CFS. It also underscores the complexity of potential interventions, which must address multiple interacting cell types and pathways within a systems-biology paradigm integrating immunology (e.g., cytokine signaling), neurology (synaptic and network dynamics), vascular physiology (cerebral perfusion and barrier integrity), and energy metabolism (mitochondrial and astrocytic support) [237]. Peripheral abnormalities in immune, muscular, or gut systems are not excluded; rather, they are understood to influence brain function via the vagus nerve, blood–brain barrier, and circulating mediators [238]. Importantly, human case–control studies in ME/CFS consistently demonstrate immune cell–specific alterations, including reduced natural killer cell cytotoxicity and transcriptional remodeling of circulating monocyte and T-cell populations, providing a concrete peripheral immune substrate that can engage the neuroimmune and neurovascular mechanisms outlined in this review.
Looking forward, ME/CFS cognitive research stands at a pivotal stage. The surge of post-COVID-19 condition cases has created a large cohort exhibiting post-viral brain-fog symptoms that overlap mechanistically with ME/CFS, accelerating biomarker discovery and mechanistic insight [239]. Shared features include post-infectious immune activation, autonomic dysfunction, cerebral hypoperfusion, and glial metabolic stress, suggesting partially overlapping neuroimmune and neurovascular mechanisms [239]. Advances in neuroimaging, such as 7-Tesla MRI, astrocyte- or microglia-specific PET tracers, and nanobody-based contrast agents, may soon permit in-vivo visualization of subtle glial and vascular changes [56]. In the near term, priority should be given to longitudinal imaging–biomarker studies that link glial or vascular measures to objective cognitive outcomes, thereby enabling rational selection of targets for early-phase intervention trials. Multi-omics combined with machine-learning approaches is beginning to stratify ME/CFS into biologically meaningful subtypes such as neuroinflammatory-dominant and autonomic-dominant phenotypes, which lays an essential foundation for future precision-medicine trials [233, 240]. Future clinical research will likely adopt combination approaches, for example pairing an anti-inflammatory agent with a vasodilator and cognitive rehabilitation, possibly within adaptive-design trial frameworks. Preventive strategies may also emerge if early immune dysregulation after infection is shown to prime microglia toward chronic activation. With sustained interdisciplinary collaboration and targeted funding, the coming decade may see ME/CFS research evolve from descriptive observation to mechanistic and interventional science. This framework yields testable predictions, including that normalization of cerebral perfusion or glial metabolic markers should precede measurable cognitive improvement, and that patient subgroups defined by dominant neurovascular or neuroimmune signatures will exhibit differential treatment responsiveness.