Metabolomic and immune alterations in long COVID patients with chronic fatigue syndrome

Abstract

Introduction: A group of SARS-CoV-2 infected individuals present lingering symptoms, defined as long COVID (LC), that may last months or years post the onset of acute disease. A portion of LC patients have symptoms similar to myalgic encephalomyelitis or chronic fatigue syndrome (ME/CFS), which results in a substantial reduction in their quality of life. A better understanding of the pathophysiology of LC, in particular, ME/CFS is urgently needed.

Methods: We identified and studied metabolites and soluble biomarkers in plasma from LC individuals mainly exhibiting ME/CFS compared to age-sex-matched recovered individuals (R) without LC, acute COVID-19 patients (A), and to SARS-CoV-2 unexposed healthy individuals (HC).

Results: Through these analyses, we identified alterations in several metabolomic pathways in LC vs other groups. Plasma metabolomics analysis showed that LC differed from the R and HC groups. Of note, the R group also exhibited a different metabolomic profile than HC. Moreover, we observed a significant elevation in the plasma pro-inflammatory biomarkers (e.g. IL-1α, IL-6, TNF-α, Flt-1, and sCD14) but the reduction in ATP in LC patients. Our results demonstrate that LC patients exhibit persistent metabolomic abnormalities 12 months after the acute COVID-19 disease. Of note, such metabolomic alterations can be observed in the R group 12 months after the acute disease. Hence, the metabolomic recovery period for infected individuals with SARS-CoV-2 might be long-lasting. In particular, we found a significant reduction in sarcosine and serine concentrations in LC patients, which was inversely correlated with depression, anxiety, and cognitive dysfunction scores.

Conclusion: Our study findings provide a comprehensive metabolomic knowledge base and other soluble biomarkers for a better understanding of the pathophysiology of LC and suggests sarcosine and serine supplementations might have potential therapeutic implications in LC patients. Finally, our study reveals that LC disproportionally affects females more than males, as evidenced by nearly 70% of our LC patients being female.

Keywords: cognitive performance; depression; sarcosine; serine; soluble CD14.

Figure 1

Demographic analysis of LC cohorts. (A) schematic of the study design. Numbers in center of diagram indicate participants in each study cohort (HC, healthy controls with no prior-SARS-CoV-2 exposure/vaccination; A, acute SARS-CoV-2 infected and ICU-admitted; LC, Long COVID; R, recovered). Outer ring indicates different studies/assays performed on patients/samples. (B) Demographic characteristics for each group displayed as ring charts for sex and age.

Figure 2

Altered metabolomic profile in SARS-CoV-2 infected individuals. (A) Partial least squares-discrimination analysis (PLS-DA) plot based on the metabolites in LC (n=30), acute COVID-19 (n=15), HC (n=15) and R (n=15). (B) Heatmap based on ANOVA using the top 100 significantly altered metabolites. The heatmap indicates the auto scaled levels of each metabolite in each sample, colored blue for decline and red for elevation as indicated on the horizontal bar. (C) Partial least squares-discrimination analysis (PLS-DA) plot based on the metabolites in LC (n=30), HC (n=15) and R (n=15). (D) Heatmap based on ANOVA using the 100 significantly altered metabolites. The heatmap indicates the auto scaled levels of each metabolite in each sample, colored blue for decline and red for elevation as indicated on the horizontal bar. (E) Volcano plots of significantly increased (red), decreased (blue) or unchanged (black) metabolites in acute A vs HCs. (F) Volcano plots of significantly increased, decreased or unchanged metabolites in R vs HC. (G) Volcano plots of significantly increased, decreased or unchanged metabolites in LC vs HC. (H) Volcano plots of significantly increased, decreased or unchanged metabolites in A vs R. (I) Volcano plots of significantly increased, decreased or unchanged metabolites in A vs LC. (J) Volcano plots of significantly increased, decreased or unchanged metabolites in LC vs R.

Figure 3

Altered metabolomic profile in LC vs HC. (A) Metabolic pathway enrichment analysis plot shows significantly altered pathways in LC vs HC. (B) The heatmap shows significantly decreased metabolites in different metabolic pathways in LC vs R. (C) The heatmap shows significantly elevated metabolites in different metabolomics pathways in LC vs R.

Figure 4

Altered metabolomic profile in LC vs R. (A) Metabolic pathway enrichment analysis plot shows significantly altered pathways in LC vs R. (B) The heatmap shows significantly decreased metabolites in different metabolic pathways in LC vs R. (C) The heatmap shows significantly elevated metabolites in different metabolomics pathways in LC vs R.

Figure 5

Altered metabolomic profile in R vs HC. (A) Metabolic pathway enrichment analysis plots of significantly altered pathways in the R vs HC group. (B) The heatmap shows significantly increased metabolites in different metabolic pathways in the R vs HC group. (C) The heatmap shows significantly decreased metabolites in different metabolic pathways in the R vs HC group.

Figure 6

Elevated levels of proinflammatory biomarkers in LC patients. (A) Cumulative data comparing the plasma IL-1α, (B) IL-6, (C) TNF-α, (D) IP-10, (E) CRP, (F) SAA, (G) IL-13, and (H) IL-1β, (I) Flt-1, and (J) soluble CD14 measured by ELISA in the plasma of R, LC, and HC groups. (K) Comparing the Anti-CaSR antibody levels in plasma samples of HC, R, and LC group. (L) Soluble CaSR levels in plasma samples of HC, R, LC and acute COVID-19 patients. (M) Cumulative data of the plasma ATP in HC, acute, R, and LC groups. Kruskal–Wallis analysis with Dunn’s multiple comparisons test. ns, not significant. Each dot represents a study subject. * p < 0.5, ** p ≤ 0.01, *** p ≤ 0.001, and **** p ≤ 0.0001.

Figure 7

The association of sarcosine and serine with clinical symptoms in LC. (A) The correlation between serine levels with anxiety score, and (B) depression score in LC patients. (C) The correlation between sarcosine levels with cognitive failure score, and (D) depression score in LC patients. Each dot represents a study subject. P values and R² were obtained by Simple linear repression analysis.