Beyond plasma signatures: reframing bioenergetic disruption in long COVID as a neurological disorder

Dear Editor,

García-Hidalgo and colleagues [1] present an elegant and technically sophisticated metabolomic analysis of survivors of critical COVID-19, identifying a multilayer plasma signature associated with persistent symptoms at 12 months. The integration of NMR- and GC/MS-derived metabolites with machine-learning feature selection is a methodological strength, and the finding of reduced circulating creatine together with elevated α-ketoglutarate offers a compelling biochemical window into post-acute sequelae. The study advances the field by framing long COVID as a disorder of disrupted bioenergetics rather than merely a residual inflammatory state.

However, an important conceptual dimension remains underdeveloped: long COVID is increasingly recognized as a predominantly neurological disorder [2], with cognitive dysfunction, central fatigue, dysautonomia, and post-exertional malaise forming its clinical core in many cohorts. In the present study, all individuals classified as long COVID exhibit abnormal cognitive scores (BC-CCI), and half meet criteria for pathological fatigue, yet the discussion treats these manifestations largely as downstream consequences of systemic metabolic disturbance. This framing risks obscuring the central nervous system as a primary locus of pathology.

Over the past three years, convergent neurobiological evidence has emerged showing that long COVID is characterized by impaired cerebral energetics. Magnetic resonance spectroscopy (MRS) studies have demonstrated reductions in brain creatine and phosphocreatine, altered ATP flux, and region-specific metabolic slowing in frontal and limbic circuits [3]. Parallel findings in post-exertional malaise and myalgic encephalomyelitis/chronic fatigue syndrome, conditions now widely viewed as overlapping with long COVID, show reduced phosphocreatine recovery rates, diminished mitochondrial oxidative capacity, and abnormal neural lactate accumulation during cognitive or physical challenge [4]. These tissue-level abnormalities provide a mechanistic bridge between subjective cognitive fatigue and objective bioenergetic failure.

Against this backdrop, the interpretation of reduced plasma creatine in the present study requires caution. Circulating creatine is not a direct proxy for intracellular brain or muscle creatine pools. The creatine-phosphocreatine system is highly compartmentalized, and plasma levels may reflect dietary intake, renal handling, or altered transport rather than impaired cellular energy buffering per se. A decrease in plasma creatine could signify increased tissue uptake, altered synthesis, or redistribution under chronic stress. Without concurrent tissue-level measurements, particularly in brain or skeletal muscle, the directionality and physiological meaning of this signal remain ambiguous.

Similarly, elevated α-ketoglutarate and fumarate are interpreted primarily as markers of systemic mitochondrial dysfunction. Yet in neural tissue, these intermediates are tightly coupled to glutamatergic neurotransmission, redox balance, and epigenetic regulation [5]. α-Ketoglutarate is not merely a TCA metabolite; it is a co-factor for dioxygenases regulating hypoxia signaling and chromatin state. In the brain, perturbations in α-ketoglutarate availability can reshape neuronal excitability and cognitive processing. The current discussion does not engage with this neurobiological context, despite cognitive impairment being universal in the long COVID group.

Another concern lies in the implicit causal leap from association to therapeutic inference. The authors propose creatine supplementation as a potential intervention. While biologically plausible [4], this recommendation rests on plasma correlations in a small, highly selected ICU cohort (n = 42). Long COVID is clinically heterogeneous, and many patients with profound neurocognitive symptoms were never hospitalized. Whether the identified metabolomic signature generalizes to the broader long COVID population, particularly those with predominantly neurological phenotypes, remains unknown. Moreover, supplementation strategies should ideally be grounded in demonstrations of tissue-level deficiency or impaired creatine kinase flux, rather than systemic concentrations alone.

Finally, the exclusive reliance on peripheral blood risks reinforcing a systemic-only model of long COVID. Plasma metabolomics captures the biochemical footprint of disease but not its anatomical substrate. In a condition increasingly defined by central network dysfunction, autonomic instability, and impaired cerebral energy metabolism [2], the absence of neuroimaging, neurometabolic, or cerebrospinal fluid correlates represents a missed opportunity for integration across biological scales.

In summary, García-Hidalgo and co-workers [1] provide important evidence that long COVID is associated with persistent disturbances in metabolic homeostasis. To fully realize the translational potential of these findings, future work should explicitly integrate the neurological nature of the syndrome and bridge plasma biomarkers with tissue-level bioenergetic failure, particularly in the brain. Long COVID is not merely a systemic metabolic disorder with neurological symptoms; for many patients, it is a neurological disease whose systemic signatures are secondary reflections of impaired cellular energy handling. Recognizing this hierarchy is essential for both mechanistic clarity and rational therapeutic development.

Author contributions

SMO: Conceptualization; Writing-original draft.

Funding

None.

Data availability

Not applicable.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

The author gives permission for the Journal to publish this work.

Competing interests

None.