The research, led by Anna C. Sanford of Tulane University School of Medicine together with Nell G. Bond, John S. Schieffelin, and colleagues, drew on an ongoing cohort study of Ebola virus disease survivors and their household contacts in Sierra Leone. The team collected plasma samples between March 2016 and April 2019 and applied untargeted metabolomics—liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS)—to survey hundreds of small molecules circulating in the blood. Their comparison groups were carefully constructed: 37 survivors with post-Ebola syndrome, split between those with musculoskeletal or gastrointestinal sequelae (18) and those with cardiopulmonary sequelae (19); 20 asymptomatic survivors who tested positive for Ebola virus glycoprotein IgG antibodies; and 20 asymptomatic household contacts who lacked these antibodies, matched as closely as possible to the survivors in age, sex, body mass index, and sample collection date.

The analytical pipeline was deliberately conservative. Peak intensities were normalized to the median and log-transformed, and variables with high relative standard deviation or near-constant values were filtered out before testing. Metabolites were considered significantly different only if they showed a fold change greater than 2.0 or less than 0.5—equivalent to an absolute log2 fold change above 1—combined with an adjusted p-value below 0.05 using the Benjamini-Hochberg false discovery rate correction. Of the 544 metabolites identified in the plasma samples, 34 met these stringent criteria when survivors with post-Ebola syndrome were compared with asymptomatic survivors. Twenty-six of these metabolites were significantly decreased in the PES group, while eight were increased. Importantly, logistic regression showed that age, sex, body mass index, and sample collection date did not predict group membership, meaning the metabolic differences could not be explained by these demographic factors.

The pattern of change was strikingly one-directional. Pathway analysis in MetaboAnalyst 6.0 revealed that the most heavily impacted downregulated pathways involved amino acid metabolism—including the taurine and hypotaurine pathways and the tyrosine, cysteine, and methionine pathways—alongside nucleotide metabolism, the tricarboxylic acid (TCA) cycle, glycolysis, and pyruvate metabolism. In other words, survivors with post-Ebola syndrome showed broad suppression of the central biochemical machinery that cells use to generate energy and build proteins and nucleic acids. A smaller set of pathways was significantly upregulated or showed mixed changes, including starch and sucrose metabolism, tryptophan metabolism, arginine and proline metabolism, nicotinate and nicotinamide metabolism, and pentose and glucuronate interconversions.

Because so many TCA cycle metabolites were affected, the researchers performed a network analysis to visualize how these compounds relate to one another. Spearman’s rank correlation analysis across the PES group revealed significant interconnections among metabolites in the TCA cycle, amino acid, fatty acid, and nucleotide metabolism pathways, and the network of TCA metabolites was visualized in Cytoscape, with node size reflecting the number of significant connections and line thickness reflecting the strength of the correlation. Notably, none of the demographic variables correlated with any metabolite in the PES group, reinforcing that the metabolic signature was independent of body composition, age, or sex. Multivariate approaches added further support: principal components analysis and partial least squares discriminant analysis (PLS-DA) separated survivors with post-Ebola syndrome from asymptomatic survivors, with the best two-component PLS-DA model achieving an accuracy of 0.86, an R2 of 0.66, and a Q2 of 0.45. Metabolites tied to the TCA cycle, short-chain fatty acid metabolism, and glucose carried the greatest weight in the model.

The team then asked whether a small panel of metabolites could serve as a diagnostic biomarker for post-Ebola syndrome. Using a discovery cohort of 39 randomly selected survivors—roughly two-thirds of the group—they trained both PLS-DA and Random Forest models, running one hundred iterations of Monte-Carlo cross-validation. The top ten most important metabolites were the same in both models: acrylate, glucose, acetate, 3-hydroxy-3-methylglutarate, malate, 2-oxo-4-hydroxy-4-carboxy-5-ureidoimidazoline, succinate, methylmalonate, threonate-1,4-lactone, and propanoate. When this fixed ten-metabolite panel was applied to an independent validation cohort of the remaining 18 survivors, which the discovery model had never seen, both models again achieved areas under the receiver operating characteristic curve above 0.8, with predictive accuracies of 0.799 for PLS-DA and 0.796 for Random Forest. The convergence of two independent machine-learning approaches on the same metabolite panel, and its successful validation on held-out data, strengthens the case that these molecules genuinely track the syndrome rather than reflecting statistical noise.

What might these metabolic shifts mean biologically? The authors point to converging evidence from other viral diseases. In cells stimulated with Ebola virus-like particles, fatty acid and amino acid metabolism are known to be reprogrammed, and fatal Ebola virus disease has previously been associated with depleted plasma free amino acids. In people living with HIV, altered amino acid metabolism and signs of mitochondrial disruption—including TCA cycle dysfunction—have been documented, and impaired mitochondrial function is a recognized hallmark of T-cell exhaustion in chronic infections such as HIV, hepatitis B, and lymphocytic choriomeningitis virus infection. The TCA cycle changes observed in survivors with post-Ebola syndrome are associated with diminished CD8+ T-cell activity in those chronic infection settings, raising the possibility that similar immunometabolic alterations persist in Ebola survivors who develop long-term sequelae. Prior work by other groups has shown that more survivors with PES mount Ebola-specific CD8+ T-cell responses than asymptomatic survivors, and metabolism is known to mediate immune cell differentiation and function, making the connection biologically plausible even though plasma measurements cannot be directly mapped onto cellular processes.

The short-chain fatty acid findings add another intriguing dimension. Acetate was increased while propanoate was decreased in the PES group, and the authors suggest that this pattern could reflect dysregulation of the gut microbiome. Reduced populations of short-chain fatty acid-producing symbionts have been reported in people with Long COVID, and short-chain fatty acids generally encourage anti-inflammatory immune responses while also modulating CD8+ T-cell activity. If the gut microbial community of Ebola survivors with post-Ebola syndrome has shifted away from these protective metabolite producers, it could provide a mechanistic bridge between infection, microbial ecology, and chronic immune dysfunction. The authors are careful, however, to note that any suggestion of viral persistence as a driver of these changes remains purely hypothetical with the current data.

The study has clear limitations that the authors acknowledge candidly. The sample size is modest, and the focus on two specific sequelae groups—musculoskeletal/gastrointestinal and cardiopulmonary—may limit generalizability to the broader spectrum of post-Ebola syndrome. No significant differences emerged when survivors with PES were compared with asymptomatic household contacts, which the authors attribute to limited statistical power, trends that did not reach significance, or possibly unique metabolic changes in asymptomatic survivors themselves. There were also no significant metabolite differences between the musculoskeletal/gastrointestinal and cardiopulmonary subgroups. Concurrent laboratory measurements such as inflammatory markers were unavailable for correlation, no acute-phase samples exist to distinguish persistent changes from those dating to the original illness, and longitudinal samples are lacking. Shared analytical features among some metabolites also limit precision, and the authors call for targeted metabolomics to resolve these ambiguities. Despite these caveats, the work establishes for the first time that Ebola survivors with post-Ebola syndrome carry a distinct and largely downregulated metabolic profile centered on energy metabolism, amino acids, nucleotides, and microbially derived short-chain fatty acids. The team plans mechanistic studies to determine how these changes arise and how they relate to immune function, and hopes to compare findings directly with post-viral syndromes following other infections such as SARS-CoV-2. More broadly, the study demonstrates the power of metabolomics to illuminate the biology of post-viral illness—a field where molecular explanations have long been sought and rarely found.

Subject of Research: Metabolic biomarkers of post-Ebola syndrome in Ebola virus disease survivors

Article Title: Metabolic basis of post-infectious sequelae after Ebola virus disease

Article References: Sanford, A. C., Bond, N. G., Ficenec, S., Osterman, C., Farkas, P., Engel, E., Gunn, B., Grant, D. S., Samuels, R., Zwezdaryk, K. J., & Schieffelin, J. S. (2026). Metabolic basis of post-infectious sequelae after Ebola virus disease. Metabolomics, 22(6), Article 170. https://doi.org/10.1007/s11306-026-02544-z

Image Credits: AI Generated

DOI: 10.1007/s11306-026-02544-z

Keywords: Ebola virus disease, post-Ebola syndrome, metabolomics, TCA cycle, short-chain fatty acids, amino acid metabolism, biomarkers, Sierra Leone, T-cell exhaustion, gut microbiome, LC-MS/MS, post-viral sequelae